Advances in Manufacturing II: Justyna Trojanowska Olaf Ciszak José Mendes Machado Ivan Pavlenko Editors

Lecture Notes in Mechanical Engineering
Justyna Trojanowska
Olaf Ciszak
José Mendes Machado
Ivan Pavlenko Editors
Advances in
Manufacturing
II
Volume 1 - Solutions for Industry 4.0
Lecture Notes in Mechanical Engineering (LNME) publishes the latest develop-
ments in Mechanical Engineering - quickly, informally and with high quality.
Original research reported in proceedings and post-proceedings represents the core
of LNME. Volumes published in LNME embrace all aspects, subfields and new
challenges of mechanical engineering. Topics in the series include:
• Engineering Design
• Machinery and Machine Elements
• Mechanical Structures and Stress Analysis
• Automotive Engineering
• Engine Technology
• Aerospace Technology and Astronautics
• Nanotechnology and Microengineering
• Control, Robotics, Mechatronics
• MEMS
• Theoretical and Applied Mechanics
• Dynamical Systems, Control
• Fluid Mechanics
• Engineering Thermodynamics, Heat and Mass Transfer
• Manufacturing
• Precision Engineering, Instrumentation, Measurement
• Materials Engineering
• Tribology and Surface Technology
To submit a proposal or request further information, please contact the Springer
Editor in your country:
China: Li Shen at li.shen@springer.com
India: Dr. Akash Chakraborty at akash.chakraborty@springernature.com
Rest of Asia, Australia, New Zealand: Swati Meherishi at
swati.meherishi@springer.com
All other countries: Dr. Leontina Di Cecco at Leontina.dicecco@springer.com
To submit a proposal for a monograph, please check our Springer Tracts in
Mechanical Engineering at http://www.springer.com/series/11693 or contact
Leontina.dicecco@springer.com
Indexed by SCOPUS. The books of the series are submitted for indexing to
Web of Science.
More information about this series at http://www.springer.com/series/11236

Justyna Trojanowska Olaf Ciszak
• •
José Mendes Machado Ivan Pavlenko

•
Editors
Advances in
Manufacturing II
Volume 1 - Solutions for Industry 4.0
123
Editors
Justyna Trojanowska Olaf Ciszak
Poznan University of Technology Poznan University of Technology
Poznan, Poland Poznan, Poland
José Mendes Machado Ivan Pavlenko

Mechanical Engineering Department Sumy State University
University of Minho Sumy, Ukraine
Braga, Portugal
ISSN 2195-4356 ISSN 2195-4364 (electronic)

ISBN 978-3-030-18714-9 ISBN 978-3-030-18715-6 (eBook)
https://doi.org/10.1007/978-3-030-18715-6
© Springer Nature Switzerland AG 2019
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, expressed or implied, with respect to the material contained
herein or for any errors or omissions that may have been made. The publisher remains neutral with regard
to jurisdictional claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface
This volume of Lecture Notes in Mechanical Engineering contains selected

papers presented at the sixth International Scientific-Technical Conference
MANUFACTURING 2019, held in Poznan, Poland, on May 19–22, 2019. The
conference was organized by the Faculty of Mechanical Engineering and
Management, Poznan University of Technology, Poland, under the scientific aus-
pices of the Committee on Machine Building and Committee on Production
Engineering of the Polish Academy of Sciences.
The aim of the conference was to present the latest achievements in mechanical
engineering and to provide an occasion for discussion and exchange of views and
opinions. The main conference topics were:
• quality engineering and management
• production engineering and management
• mechanical engineering
• metrology and measurement systems
• solutions for Industry 4.0.
The organizers received 293 contributions from 36 countries around the world.
After a thorough peer review process, the committee accepted 167 papers for
conference proceedings prepared by 491 authors from 23 countries (acceptance rate
around 57%). Extended versions of selected best papers will be published in the
following journals: Flexible Services and Manufacturing Journal, Research in
Engineering Design, Management and Production Engineering Review and
Archives of Mechanical Technology and Materials.
The book Advances in Manufacturing II is organized into five volumes that
correspond with the main conference topics mentioned above.
Advances in Manufacturing II - Volume 1 - Solutions for Industry 4.0
contains current research on the Industry 4.0 concept and aims to increase the
efficiency of production processes. The use of tools supporting the concept in both
mechanical and biomedical engineering is presented. A separate chapter is devoted
to the analysis of the latest trends in robotics and describes the use of industrial
robots cooperating with humans in a smart factory. The book also includes issues
v
vi Preface
related to data acquisition and decision support using artificial neural networks. The
book consists of 37 chapters, prepared by 100 authors from 14 countries.
We would like to thank the members of the International Program Committee for
their hard work during the review process.
We acknowledge all that contributed to the staging of MANUFACTURING
2019: authors, committees, and sponsors. Their involvement and hard work were
crucial to the success of the MANUFACTURING 2019 conference.
May 2019 Justyna Trojanowska

Olaf Ciszak
José Mendes Machado
Ivan Pavlenko
Organization
Steering Committee
General Chair
Adam Hamrol Poznan University of Technology, Poland
Chairs
Olaf Ciszak Poznan University of Technology, Poland

Stanisław Legutko Poznan University of Technology, Poland
Scientific Committee
Stanisław Adamczak, Poland Jens J. Dahlgaard, Sweden

Michal Balog, Slovakia María de los Angeles Cervantes Rosas,
Zbigniew Banaszak, Poland Mexico
Myriam Elena Baron, Argentina Andrzej Demenko, Poland
Stefan Berczyński, Poland Magdalena Diering, Poland
Johan Berglund, Sweden Ewa Dostatni, Poland
Wojciech Bonenberg, Poland Jan Duda, Poland
Christopher A. Brown, USA Davor Dujak, Croatia
Anna Burduk, Poland Milan Edl, Czech Republic
Somnath Chattopadhyaya, India Sabahudin Ekinovic, Bosnia
Shin-Guang Chen, Taiwan and Herzegovina
Danut Chira, Romania Mosè Gallo, Italy
Edward Chlebus, Poland Bartosz Gapiński, Poland
Damir Ciglar, Croatia Józef Gawlik, Poland
Marcela Contreras, Mexico Hans Georg Gemuenden, Norway
Nadežda Cuboňová, Slovakia Boštjan Gomišček, UEA
vii
viii Organization
Marta Grabowska, Poland Alejandro Pereira Dominguez, Spain

Wit Grzesik, Poland Marko Periša, Croatia
Michal Hatala, Slovakia Emilio Picasso, Argentina
Sandra Heffernan, New Zealand Jan Pitel, Slovakia
Christoph Herrmann, Germany Alla Polyanska, Ukraine
Ivan Hudec, Slovakia Włodzimierz Przybylski, Poland
Viatlii Ivanov, Ukraine Luis Paulo Reis, Portugal
Andrzej Jardzioch, Poland Álvaro Rocha, Portugal
Mieczyslaw Jurczyk, Poland Rajkumar Roy, UK
Wojciech Kacalak, Poland Iwan Samardzic, Croatia
Lyudmila Kalafatova, Ukraine Krzysztof Santarek, Poland
Anna Karwasz, Poland Jarosław Sęp, Poland
Mourad Keddam, Algeria Bożena Skołud, Poland
Sławomir Kłos, Poland Jerzy Sładek, Poland
Ryszard Knosala, Poland Roman Staniek, Poland
Janusz Kowal, Poland Beata Starzyńska, Poland
Drazan Kozak, Croatia Tomasz Sterzyński, Poland
Agnieszka Kujawińska, Poland Tomasz Stręk, Poland
Janos Kundrak, Hungary Antun Stoić, Croatia
Maciej Kupczyk, Poland Manuel Francisco Suarez Barraza,
Ivan Kuric, Slovakia Mexico
Oleksandr Liaposhchenko, Ukraine Marek Szostak, Poland
Piotr Łebkowski, Poland Rafał Talar, Poland
José Mendes Machado, Portugal Franciszek Tomaszewski, Poland
Aleksandar Makedonski, Bulgaria María Estela Torres Jaquez, Mexico
Ilija Mamuzic, Croatia Justyna Trojanowska, Poland
Krzysztof Marchelek, Poland Stefan Trzcieliński, Poland
Tadeusz Markowski, Poland Maria Leonilde R. Varela, Portugal
Edison Perozo Martinez, Colombia Sachin D. Waigaonkar, India
Thomas Mathia, France Edmund Weiss, Poland
Józef Matuszek, Poland Michał Wieczorowski, Poland
Adam Mazurkiewicz, Poland Ralf Woll, Germany
Andrzej Milecki, Poland Magdalena Wyrwicka, Poland
Mirosław Pajor, Poland Jozef Zajac, Slovakia
Ivan Pavlenko, Ukraine Jan Żurek, Poland
Dragan Perakovic, Croatia
Program Committee
Available on http://manufacturing.put.poznan.pl/en/.
Organization ix
Special Sessions
Collaborative Manufacturing and Management in the Context of Industry 4.0
Special Session Organizing Committee
Leonilde Varela University of Minho, Portugal

Justyna Trojanowska Poznan University of Technology, Poland
Vijaya Kumar Manupati Mechanical Engineering Department,
NIT Warangal
José Machado University of Minho, Portugal
Eric Costa Solent University, UK
Sara Bragança Solent University, UK
Intelligent Manufacturing Systems

Ivan Pavlenko Sumy State University, Ukraine

Sławomir Luściński Kielce University of Technology, Poland
Tooling and Fixtures: Design, Optimization, Verification

Vitalii Ivanov Sumy State University, Ukraine

Yiming Rong Southern University of Science and Technology,
China
Advanced Manufacturing Technologies

Jozef Jurko TU Košice, Slovak Republic

Michal Balog TU Košice, Slovak Republic
Tadeusz E. Zaborowski TU Poznań, Poland
x Organization
The Changing Face of Production Engineering and Management

in a Contemporary Business Landscape
Damjan Maletič University of Maribor, Faculty of Organizational

Sciences, Enterprise engineering Laboratory,
Slovenia
Matjaž Maletič University of Maribor, Faculty of Organizational
Slovenia
Tomaž Kern University of Maribor, Faculty of Organizational
Slovenia
Enabling Tools and Education for Industry 4.0

Dorota Stadnicka Politechnika Rzeszowska, Poland

Dario Antonelli Politecnico di Torino, Italy
Katarzyna Antosz Politechnika Rzeszowska, Poland
Staff for the Industry of the Future

Magdalena Wyrwicka Poznan University of Technology, Poland

Anna Vaňová Faculty of Economics, Matej Bel University,
Slovakia
Maciej Szafrański Poznan University of Technology, Poland
Magdalena Poznan University of Technology, Poland
Graczyk-Kucharska
Advances in Manufacturing, Properties, and Surface Integrity

of Construction Materials
Szymon Wojciechowski Poznan University of Technology, Poland

Grzegorz M. Królczyk Opole University of Technology, Poland
Sergei Hloch Technical University of Kosice, Slovakia
Organization xi
Materials Engineering
Monika Dobrzyńska-Mizera Poznan University of Technology, Poland

Monika Knitter Institute of Materials Technology, Poznan
University of Technology, Poland
Robert Sika Institute of Materials Technology, Poznan
Dariusz Bartkowski Institute of Materials Technology, Poznan
Waldemar Matysiak Institute of Materials Technology, Poznan
Anna Zawadzka Institute of Materials Technology, Poznan
Advanced Mechanics of Systems, Materials and Structures

Hubert Jopek Institute of Applied Mechanics, Poznan

Paweł Fritzkowski Institute of Applied Mechanics, Poznan
Jakub Grabski Institute of Applied Mechanics, Poznan
Krzysztof Sowiński Institute of Applied Mechanics, Poznan
Agata Matuszewska Institute of Applied Mechanics, Poznan
Virtual and Augmented Reality in Manufacturing

Filip Górski Poznan University of Technology, Poland

Paweł Buń Poznan University of Technology, Poland
Damian Grajewski Poznan University of Technology, Poland
Jorge Martin-Gutierrez Universidad de la Laguna, Spain
Letizia Neira Universidad Autónoma de Nuevo León, Mexico
Eduardo Gonzalez Mendivil Tecnologico de Monterrey, Mexico
Contents
Industry 4.0 Concept

Digitization and Intangible Assets in Manufacturing Industries . . . . . . . 3
G. Scott Erickson and Helen N. Rothberg
Human-Organization Relation in the Perspective of Industry 4.0 . . . . . . 14
Waldemar Jędrzejczyk
Edge Computing Technologies as a Crucial Factor of Successful
Industry 4.0 Growth. The Case of Live Video Data Streaming . . . . . . . 25
Joanna Gąbka
Model Proposal to Evaluate the Quality of a Production Planning
and Control Software in an Industrial Context . . . . . . . . . . . . . . . . . . . 38
Ricardo M. P. Gonçalves, Maria L. R. Varela, Ana M. Madureira,
Goran D. Putnik, and Jose Machado
Increasing the Production Productivity with Artificial Bee Colony
Optimisation Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Danijela Pezer
Comparison of the NC Data Preparation Methods
for Drawingless Production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Peter Kostal, Andrea Mudriková, Daynier R. D. Sobrino, David Michal,
and Simon Lecky
Characteristics of Resources as a Determinant of Implementation
of the Physical Internet Concept in Supply Chains . . . . . . . . . . . . . . . . 72
Michał Adamczak, Łukasz Hadaś, Agnieszka Stachowiak,
Roman Domański, and Piotr Cyplik
Model of a Knowledge Management for System Integrator(s)
of Cyber-Physical Production Systems (CPPS) . . . . . . . . . . . . . . . . . . . . 92
Justyna Patalas-Maliszewska and Nadine Schlueter
xiii
xiv Contents
Knowledge Management in Open Industrial Virtual

Reality Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
Filip Górski, Paweł Buń, Przemysław Zawadzki,
and Radosław Wichniarek
The Automated Acquisition of Expert Knowledge Using a Service
Department as an Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Justyna Patalas-Maliszewska, Adam Dudek, and Sławomir Kłos
Virtual Simulation of Machine Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
Damian Grajewski, Filip Górski, and Zoran Pandilov
Cyber Sickness in Industrial Virtual Reality Training . . . . . . . . . . . . . . 137
Magdalena Żukowska, Paweł Buń, Filip Górski, and Beata Starzyńska
Does Metal Additive Manufacturing in Industry 4.0 Reinforce
the Role of Substractive Machining? . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
Marcin A. Królikowski and Marta B. Krawczyk
Ensuring Vibration Reliability of Turbopump Units Using Artificial
Neural Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
Ivan Pavlenko, Vitalii Ivanov, Ivan Kuric, Oleksandr Gusak,
and Oleksandr Liaposhchenko
Interaction in Project Management Approach Within Industry 4.0 . . . . 176
Mehmet Cakmakci
Applying Scrum in New Product Development Process . . . . . . . . . . . . . 190
Jan Betta, Tomasz Chlebus, Dorota Kuchta, and Agnieszka Skomra
Smart Industry Solutions

Drivers Impacting Cobots Adoption in Manufacturing Context:
A Qualitative Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
Ana C. Simões, António Lucas Soares, and Ana C. Barros
Control of the 6-Axis Robot Using a Brain-Computer Interface
Based on Steady State Visually Evoked Potential (SSVEP) . . . . . . . . . . 213
Arkadiusz Kubacki and Andrzej Milecki
An Innovative Approach of Industrial Robot Programming Using
Virtual Reality for the Design of Production Systems Layout . . . . . . . . 223
Radovan Holubek, Roman Ružarovský, and Daynier R. D. Sobrino
Localization of Passive RFID Tags by Small Cartesian Robot . . . . . . . . 236
Tomasz Kapłon and Andrzej Milecki
Analysis of Cooperative Industrial Task Execution by Mobile
and Manipulator Robots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
Khurshid Aliev and Dario Antonelli
Contents xv
Simulation Modeling of Assembly Processes

for Digital Manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
Sławomir Kłos and Justyna Patalas-Maliszewska
An Indicators Framework for Sustainability Assessment of RFID
Systems in Manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
Bartłomiej Gładysz and Aldona Kluczek
Automated Training of Convolutional Networks by Virtual 3D Models
for Parts Recognition in Assembly Process . . . . . . . . . . . . . . . . . . . . . . . 287
Kamil Židek, Peter Lazorík, Ján Piteľ, Ivan Pavlenko,
and Alexander Hošovský
3D Printed Hand Exoskeleton - Own Concept . . . . . . . . . . . . . . . . . . . . 298
Jakub Kopowski, Izabela Rojek, Dariusz Mikołajewski, and Marek Macko
Mechanical Analysis of Leaf Springs Realized
by Additive Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
Ciprian I. Rizescu, Daniel Besnea, Dana Rizescu, Edgar Moraru,
and Victor Constantin
An Introductive Training for Welding Workshop: A Biometric
Evaluation Using Virtual Reality Scenes to Improve Practice . . . . . . . . 319
Francisco Torres-Guerrero, Leticia Neira-Tovar, and Luis Torres-Treviño
Simulation of Flexible Manufacturing Systems as an Element
of Education Towards Industry 4.0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
Rafał Kluz and Katarzyna Antosz
VR and AR in Lean Manufacturing Classes . . . . . . . . . . . . . . . . . . . . . 342
Paweł Buń, Justyna Trojanowska, and Paulina Rewers
A Case Study of a Virtual Training Environment . . . . . . . . . . . . . . . . . 352
Vasiliki Liagkou and Chrysostomos Stylios
Virtual and Augmented Reality for Configuring, Promoting
and Servicing Household Appliances . . . . . . . . . . . . . . . . . . . . . . . . . . . 368
Krzysztof Walczak, Jakub Flotyński, Dominik Strugała, Dariusz Rumiński,
Mikołaj Maik, Anna Englert, and Tomasz Jenek
The Effectiveness of Multimodal Sensory Feedback on VR Users’
Behavior in an L-Collision Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381
Sumin Kim, Krzysztof Izdebski, and Peter König
Intelligent Systems in the Railway Freight Management . . . . . . . . . . . . 390
Michal Balog, Hanna Sokhatska, and Angelina Iakovets
LiDAR Based System for Tracking Loader Crane Operator . . . . . . . . . 406
Karol Miądlicki and Mateusz Saków
xvi Contents
Artificial Neural Network-Supported Selection of Materials

in Ecodesign . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422
Izabela Rojek and Ewa Dostatni
The Application of Artificial Neural Network in Diagnosis
of Sleep Apnea Syndrome . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 432
Renata Ferduła, Tomasz Walczak, and Szczepan Cofta
Artificial Hand Controlled by a Glove with a Force Feedback . . . . . . . . 444
Marcin Białek, Dominik Rybarczyk, Andrzej Milecki, and Patryk Nowak
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 457
Industry 4.0 Concept
Digitization and Intangible Assets
in Manufacturing Industries
G. Scott Erickson1(&) and Helen N. Rothberg2

1
Ithaca College, Ithaca, USA
gerickson@ithaca.edu
2
Marist College, Poughkeepsie, USA
Abstract. Paper analyzes prospects for digitization in select manufacturing

industries, those specifically studied in a recent consulting report. Beyond the
conclusions of that report, this study looks more deeply into how digitization
will be employed to enhance efficiency, the worker-data system interface and
other expected outcomes. Through established metrics for assessing enterprise
holdings of intangible assets such as data, knowledge, and intelligence, this
paper more precisely identifies where and how digitization might be employed
as well as the expected level of success. Overall, these industries are likely to see
more incremental improvements in processes, mainly from successfully
employing big data systems. Employee knowledge will likely be enhanced by
digitization only slightly and mainly at the operational level. Major new insights
from intelligence, especially at higher levels of these manufacturers, will likely
be rare.
Keywords: Digitization Big data Knowledge management Intelligence

Intangible assets
1 Introduction
Over the years, considerable interest has focused on the successful exploitation of
tangible assets by firms. More recently, this attention has also included intangible
assets. But even intangible assets have changed in the last couple years, with firms not
only investing in intellectual property and employee know-how but also in big data
systems. A deeper understanding of all these assets is important as industries move into
a new digital world.
One important area of interest is what capabilities firms need to compete in the
knowledge economy. As a consequence, the fields of knowledge management and
intellectual capital have grown rapidly over the last few decades. More recently, the
advent of big data has drawn similar attention, bringing intangibles not necessarily
considered “knowledge” into the discussion. Thus, we ask the questions of how much
effective use of data and information can add to a company’s success? How about
effective management of knowledge assets and/or the ability to draw intelligence
insights from other intangibles?
A recent McKinsey report on the readiness of heavy manufacturing industries to
employ digitization is one contribution to this understanding. In this paper, we take the

J. Trojanowska et al. (Eds.): Advances in Manufacturing II - Volume 1, LNME, pp. 3–13, 2019.
https://doi.org/10.1007/978-3-030-18715-6_1
4 G. S. Erickson and H. N. Rothberg
McKinsey results and subject them to further analysis on the basis of specific types of
intangible assets. In particular, we use metrics for firm holdings of big data, knowledge
(explicit and tacit), and intelligence to assess both industry averages and success rates
in exploiting these intangibles. From these results, we can better assess the potential for
digitization in these industries and the more specific ways in which digital and other
intangible assets can be used to improve manufacturing performance.
2 Research Problem
One of the earliest themes of research into better exploiting employee know-how or
knowledge in firms was the distinction between tacit and explicit knowledge. Although
chiefly associated with the knowledge management (KM) discipline, the distinction
was important to intellectual capital (IC) studies as well. Identifying and assessing the
knowledge in employees’ heads (IC) as well as then better leveraging that knowledge
through sharing and learning (KM) both depended on understanding the differences.
Tacit and explicit knowledge were first discussed in sociology [1] and added to
more business-specific applications with the development of knowledge management
interest in the early 1990’s [2]. Explicit knowledge is employee learnings or know-how
that is easier to communicate, explain, or codify. As a result, explicit knowledge is
more readily harvested from individual employees and converted into sharable formats
such as procedures or process documents. Tacit knowledge, on the other hand, is harder
to identify, explain or communicate to others, and capture in documentation. It may be
unique to the employee and hard or even impossible to share. Tacit knowledge is much
more personal.
As our understanding of the different types of knowledge grew, so did practices to
better manage them. Many substantial KM installations around the turn of the century
included considerable information technology components, elements lending them-
selves well to knowledge assets more on the explicit side [3, 4]. Tacit KM solutions
tended to be more person-to-person, often a necessity in sharing such individualized
knowledge. Typically much harder to scale across the firm, common tacit KM tech-
niques include mentoring, communities of practice, storytelling, and similar methods of
sharing [5].
Best practice in the field, then, developed into identifying the nature of the
knowledge assets within the firm and installing the most appropriate solutions (IT or
person-to-person). Tacit to explicit is now more commonly seen as a continuum, with
few purely tacit knowledge assets and few purely explicit knowledge assets, making
the choice of KM system even more complex and important. If anything, we now see
knowledge management as a critical and difficult aspect of overseeing the firm, from
the strategic level down to specific human resource and information technology
choices.
Even so, the stakes have risen even further in recent years with the advent of
interest in big data and business intelligence. Knowledge management scholarship has
always recognized the presence of data and information as potentially valuable
intangible assets in addition to knowledge [6, 7]. Often considered more of a source of
learning that could be turned into more valuable personal knowledge, data and
Digitization and Intangible Assets in Manufacturing Industries 5
information were defined and explicitly noted within the field. This perspective came
out of information technology (IT) scholarship, an important precursor to the KM/IC
disciplines, and IT’s DIKW (data, information, knowledge, wisdom) hierarchy for-
mally defined each concept, passing that structure along to other disciplines [8].
Now that big data and the related area of business analytics/intelligence are bur-
geoning areas of interest, data and analysis are starting to creep into the KM/IC dis-
cussion. In particular, big data is often treated as synonymous with knowledge,
especially explicit knowledge, in an increasing number of papers. There are similarities
but we believe the truth goes a bit deeper.
Kurtz and Snowden [9] proposed their “sensemaking” framework several years
ago, noting the distinction between intangibles collected and distributed through a
centralized hub (as with KM systems) and those shared across the far reaches of
network. Even though not intended as a hierarchy, their framework conceptually leads
to something similar to DIKW, with intangibles running from data/information through
explicit knowledge (both easily shared through systems) on to tacit knowledge (more
personal sharing required) and then to insight (creative processes so personal that they
may be unsharable) [10].
From there, it’s an easy step to a revised DIKW with data/information (collected
bytes), explicit knowledge (structured knowledge), tacit knowledge (unstructured
knowledge), and intelligence (creative insights) [11]. And if there is strategic benefit to
assessing a firm’s tacit vs. explicit knowledge assets, there is an even greater imperative
to understand how well an organization manages the full range of intangible assets and
how that compares to industry competitive requirements.
Recent work has begun to tease out these differences. Just as industries can be
assessed in terms of the presence or importance of tacit vs. explicit knowledge, so they
can be evaluated on the basis of the presence of the full range of intangible assets.
Indeed, the amount of big data can be easily tracked by sector, looking at measures like
data storage per firm [12]. So those areas where big data is apparent and consequential
are quite clear.
Similarly, metrics on knowledge (or intangibles more broadly) can help to identify
conditions related to other assets. Knowledge, in general, can be estimated in multiple
ways [13]. Single firms can often do an audit, adding up knowledge resources to arrive
at a full firm assessment. To conduct such comparisons across multiple firms is more
difficult, but accepted methods using inputs such as financial results are available.
Tobin’s q, for example, has been used as a proxy for knowledge as it calculates
intangible assets as a residual from comparisons of company value and tangible assets
[14]. Without other evidence, we often take the Tobin’s q as evidence of explicit
knowledge as it is capable of scaling enough to show up on financial reports. When
combined with intelligence indicators, discussed momentarily, Tobin’s q may also
suggest tacit knowledge is present and important [15].
Finally, insight/intelligence can also be assessed. As indicated by the revised
intangibles hierarchy, intelligence is at a level beyond tacit knowledge. Consequently,
it is best understood as being even more personal, even harder to explain or share with
others. In innovation, the creative spark that drives invention somewhat encapsulates
this perspective, that it is individual genius deriving from the ability to see new insights
from a range of data or knowledge inputs. And that individual genius is extremely hard
to teach others. If it can be done at all, it would be in one-on-one circumstances such as

when doctoral researchers train others in their labs.
So how do we assess insight? In our case, we have identified one type of intelli-
gence activity, competitive intelligence. By tracking the size and proficiency of intel-
ligence operations in firms, one can gain a sense of how widely and intensely
competitive intelligence takes place in those particular industries [16]. Firms/industries
with insight/intelligence capabilities in one area (competitive intelligence) are more
likely to also have competencies in other areas as well.
Combining all the indicators, one can get a sense of industries with:
• Substantial big data (high data metric)
• Substantial explicit knowledge (high Tobin’s q)
• Substantial tacit knowledge (high Tobin’s q and high intelligence), and
• Substantial intelligence (high intelligence).
These different circumstances suggest different strategies for managing the full
range of intangibles [15]. In some cases, only big data may be present and important to
competitiveness. Previous research, for example, has shown utilities to often have only
big data intangibles to any significant degree. In others, all four levels of intangibles
may be critical (e.g. pharmaceuticals or software). Or combinations. Financial services,
for example, typically have the highest levels of big data by industry and high intel-
ligence activity but very little knowledge or any type.
The main point is that we have some indicators allowing us to assess the com-
petitive conditions in specific industries according to intangible assets holdings and
management. That allows for some interesting analysis along those lines.
3 Methodology
This paper looks specifically at what such intangibles analysis can tell us about the
prospects for heavy manufacturing industries identified as having potential for cap-
turing digital value. McKinsey [17] recently issued a report “Mapping Heavy Indus-
tries’ Digital-Manufacturing Opportunities”, classifying various industries according to
their perceived potential to derive value from digital innovations as technology
matures. These opportunities were specific initiatives such as data analytics, a digital
workforce, asset network value maximization, and robotics/cobotics. But the main
message was that manufacturing industries and resident firms had different levels of
preparation for adopting digital strategies and, consequently, for success.
Would a deeper look at intangibles capabilities add something to this analysis? Is a
digital capability dependent on already being able to utilize big data? Or explicit
knowledge? Or intelligence? Where and how should decision makers in these indus-
tries make digital investments?
More specifically, the literature and big data practice strongly suggest that different
situations exist for the collection, processing, and analysis of data/information and
knowledge. In the area of big data, we can make a distinction between monitoring data
and analyzing data [18, 19]. In the former case, firms often identify key performance
indicators (KPI’s), tracking them on dashboards, and then taking action if performance
moves outside designated tolerances. Indeed, the actions can be left to algorithms
created by decision makers or even developed through artificial intelligence. But while
the data may be processed into a form allowing easy review and tracking, they aren’t
really analyzed for new insights.
Deeper study for new insights is more the data analytics piece of the puzzle, often
in the area of predictive analytics. In these areas, more insight or creativity are required
as analysts study the data to uncover non-obvious correlations or predictors. The
process, in fact, unearths new knowledge or, as referred to in the literature review,
intelligence. Consequently, the process and outcomes are quite different from the
monitoring activities associated with basic big data (or the sharing activities related to
knowledge management). Big data analytics requires considerably more in terms of big
jumps in human discovery rather than incremental improvements in human or machine
operations.
Consequently, we look specifically at the intangibles metrics for the key industries
identified in the McKinsey report. We do so for specific time periods for all metrics but
extend the analysis for the knowledge (Tobin’s q) metrics available for this paper. In
doing so, we are able to assess differences in intangible asset holdings and industry
conditions in these manufacturing environments, leading to a deeper analysis of what
might be required for different aspects of capturing digital value in manufacturing.
As noted, data are available on both big data holdings [12] and competitive intelli-
gence activity. The big data figures are fairly general, just breaking down data holdings by
process manufacturers vs. product manufacturers, but they are still helpful in under-
standing the presence and impact of data in those sectors. For intelligence, we employ a
data base constructed from an ongoing survey of CI professionals by consultancy Fuld &
Co. for the years 2004–2009, including almost 1,000 firms worldwide [16].
We also draw from two databases constructed to assess knowledge assets, utilizing
a modified Tobin’s q, both market capitalization to shareholders’ equity (assets less
liabilities) and market cap to assets. The two metrics are generally similar in their
results but the latter does remove debt as a major factor, an adjustment that matters in
some industries. One database covers the same time period as the other metrics, the
other updates the results.
More specifically, data are drawn from financial reports of public companies listed
on North American exchanges. For the period 2004–2009, data were collected from
I/B/E/S. The full methodology has been previously published [16] but basically
includes all annual reports with earnings above $1 billion. This encompasses almost
2,000 firms and over 7,000 entries (firms go in and out of the database as earnings go
above or below the threshold and results are also impacted by merger and acquisition
activity). In addition, for the purposes of this paper, we have collected data in the same
manner but utilizing Compustat for the years 2010–2014. The update provides some
additional insight into organizational results pertaining to knowledge resources
according to the modified Tobin’s q metric.
4 Results
To frame the results, consider again the recent McKinsey report on digital-
manufacturing opportunities in heavy industry [17]. The study specifically looked at
the following industries:
• Mining extraction
• Mining beneficiation
• Chemicals
• Petrochemicals
• Refining
• Pulp & paper
• Steel
Overall, each industry was evaluated for its “digital maturity”, ranging through four
stages, from programmable digital logic control to digital control systems, advanced
process control to artificial intelligence. The first three industries fall into the second
group and the final four into the third group, with refining closest to incorporating
artificial intelligence into its set of competencies. Industries were also rated in terms of
specific outcomes related to digital opportunities. In this case, enhanced asset perfor-
mance (just about all high), digital workforce (all low except for steel), asset network
coordination (all low to medium), and robotics (all low except steel (medium) and
mining extraction (high).
What does this scenario suggest in relation to intangible assets and their applica-
tion? Almost certainly, progress toward digitization is focused more on the big data
monitoring and explicit knowledge end of things rather than tacit knowledge or
intelligence. Efficiencies and performance are at the heart of what the consultants see
happening in these industries. Workforce implications, as noted, are mentioned as an
indicator, and those have more of a knowledge aspect to them as employees apply data
and learn, but even these seem much more likely to be incremental, explicit knowledge
learnings and sharing rather than more insightful tacit. And even though some of these
industries are approaching artificial intelligence capabilities, these are also more likely
to be more incremental learnings than truly dramatic intelligence insights more com-
monly associated with the terminology.
At least those are preliminary expectations based on the report’s conclusions. What
do additional data say?
Results are presented in Table 1, including the industries noted in the McKinsey
report and their relevant intangible metrics. The nature of the collected data requires
some adjustments to industry categories. The underlying databases are organized by
reported Standard Industrial Classification (SIC). The SIC has actually been superseded
by a newer North American Industrial Classification System but companies still list
both in their financial reports and applying SIC allows us to make more comparisons
with older databases, when appropriate.
The industries in the table, then, include mining (SIC 1, combining the two cate-
gories in the McKinsey study), chemicals (SIC 28, combining chemicals and petro-
chemicals), refining (SIC 29), pulp & paper (SIC 26), and metals (SIC 33, including
Table 1. Intangible metrics for select heavy industries

Industry Market cap/book Market cap/assets CI Stored data/firm
index (terabytes)
2005– 2010– 2005– 2010– 2005– 2011
2009 2014 2009 2014 2009
Mining 2.43 2.09 1.02 0.89 20 Discrete = 967
(n = 536) (n = 565) Process = 831
Chemicals 2.99 4.25 1.39 1.57 142
(n = 466) (n = 437)
Refining 2.69 1.90 1.06 0.78 9
(n = 106) (n = 100)
Pulp & 1.77 2.74 0.63 0.77 13
paper (n = 123) (n = 100)
Metals 2.50 1.77 0.81 0.68 8
(n = 144) (n = 105)
Global 2.68 3.61 1.02 1.06
mean
steel). The table then tracks metrics for each industry. Initially, modified Tobin’s q are
listed for two timespans, 2005–2009 and 2010–2014. For each period, two versions are
included. The traditional Tobin’s q is market capitalization to replacement value of
assets. As replacement value can be hard to obtain, a common variation is market cap
to book value (Market/book in the table). As book value takes into account both assets
and liabilities, debt can dramatically impact the measure. So we have commonly also
included market cap to assets (Market/assets in the table) which provides some idea of
the ability of the firm to generate value from resident assets irrespective of who actually
owns them.
The table also includes data on competitive intelligence (CI) activity in each
industry. As noted, these data come from a Fuld & Co. database reflecting both the
number of professionals, by firm, in each industry and their relative proficiency of their
CI operation. The index is a mix of both numbers, establishing an intelligence capa-
bility—an ability to analyze data, information, and knowledge inputs and find patterns
should be transferable, so intelligence in one area (CI) should be transferable to others
(business intelligence, marketing intelligence, etc.). More recent data (from both Fuld
& Co. and the Society of Competitive Intelligence Professionals), not yet fully pro-
cessed, shows similar results.
The table also includes the big data metrics from the McKinsey Global Services
study mentioned earlier [12]. The level of detail only drills down to a distinction
between discrete vs. process manufacturing. Those numbers reflect the average stored
data per firm in each industry group, with product somewhat higher. These numbers are
above average for a full database somewhat skewed by huge amounts of data held by
financial services and entertainment firms. For our purposes, however, the data do show
considerable levels of big data in these heavy industries. Most are more process-
oriented (continuous or batch production) than discrete (production of individual items)
but the results do agree with the newer McKinsey study that these industries are already
collected and applying big data. They are well-positioned to grow their digital
capabilities.
In terms of the Tobin’s q knowledge metrics, we can notice several interesting
patterns. Initially, the two metrics do generally agree, at least in terms of whether the
particular industry group is above or below average and whether each group has
increased or decreased its apparent average knowledge assets over the time span. All of
the industries have declined from the first measure to the second except for chemicals
and pulp/paper. All are below average in terms of apparent knowledge assets by firm,
again except for chemicals. As noted earlier, the chemicals industry results are skewed
by some very high observations coming out of pharmaceuticals, especially biologicals
as well as noticeably high results from consumer goods such as health and beauty.
But the vast majority of firms in all of these industries are at or below average in
terms of the perceived knowledge assets. Based on what we’ve seen in other studies in
other industries, these assets are likely explicit as they do scale enough to show up in
the metric (they are easily sharable across employees) and we don’t see much in the
area of intelligence, a metric implying tacit knowledge is also present in the Tobin’s q
metric. What is apparent is explicit knowledge, obtained when employees learn to
improve processes, particularly repetitive processes with more commodity type goods.
With a more flexible processes and differentiated goods, both the knowledge metric and
intelligence metric would be higher (as with the pharmaceuticals in this study). As a
result, the evidence suggests incremental, sharable knowledge is being obtained and
applied by the workforce. Explicit knowledge is there, in all industries and should be
effectively managed while substantive tacit knowledge is only in the select sectors
noted.
That conclusion again aligns with the recent McKinsey study as feedback through
big data systems, when acted upon by operators and line employees, is likely to be
incremental improvements in their area of responsibility. The firms are apparently
already aligning the workforce with digital capabilities, so as those capabilities
improve, so should their performance.
Regarding the intelligence metric, by itself it indicates the amount of competitive
intelligence (CI) activity in the industry. It is not a per firm measure as the data set is a
sample, not a complete census like the financial reports. The index combines the
number of identified industry operatives along with their specific firm’s level of pro-
ficiency. In the global data set, less than double figures is a low score (and there are
numerous industries with no apparent activity, a score of zero). The median industry
score is somewhere in the 10–25 range, and huge outliers exist in especially highly
competitive, active industries.
In the case of these industries, almost all at below the median for intelligence
activity. Mining is a little higher than the others but much of that may be due to the
prospecting function (identifying and securing high-potential sites) more than opera-
tions. Once again, the obvious outlier is chemicals but, once again, the underlying data
shows that almost all of the CI metric’s value is located in pharmaceuticals. The rest of
the industries within the broader chemicals category look more like the rest of the
sample.
What these patterns suggest is, again, little tacit knowledge obtained and shared by
employees. These are more personal learnings about job performance, harder to explain
or share. Consequently, they may be more substantial on an employee-by-employee
basis but are harder to scale across the enterprise through sharing to other workers. The
generally low intelligence metric also illustrates few new insights or creative ideas
coming out of these firms in these industries. There is very little strikingly new under
the sun.
The pharmaceuticals industry is the exception verifying some of these conclusions.
We don’t have a separate big data metric, but that figure remains fairly substantial, as in
the other industries. What we do have a much higher knowledge metric and an
extremely high intelligence score. This suggest not only the presence of big data but
also both explicit and tacit knowledge as well as intelligence. The knowledge metric, as
noted, is high, a condition that doesn’t happen without substantial explicit knowledge.
And this makes sense as we know that pharmaceutical supply chains, production, and
distribution are all closely monitored and subject to constant attempts to further opti-
mize (and document) all processes.
The intelligence metric, when combined with the high Tobin’s q, also suggests tacit
knowledge is present. Harder to share but sometimes more substantial, tacit learnings
also take place in firms like these that actually experiment on how to increase and
optimize efficiency, potency, and overall quality. Tacit may also show up in personal
relationships such as those found in the sales function and interactions with regulators.
The intelligence metric also indicates intelligence, of course, in this case the rampant CI
amongst pharmaceutical firms. Similar capabilities would also show up in R&D and
marketing strategies, functions that, again, we know are prominent in this industry but
perhaps not as much in the other manufacturers discussed.
Finally, how do these results relate to the conclusions of the recent McKinsey
report? The intangible asset data presented here largely confirm the positive digital
future of heavy manufacturing industries. But these data also provide some additional
detail and context. Big data are available to firms in these industries, providing
opportunities for asset performance enhancement, asset network valuation, and
robotics. Beyond these applications of data, the workforces should be able to improve
their performance through access to the same digital assets. The various metrics on
intangibles (data, explicit knowledge, tacit knowledge, intelligence) support these
conclusions. But they also suggest that improvements in performance will be incre-
mental. Big data will be monitored, with adjustments made, perhaps through algo-
rithms, as indicated. Similarly, workers will learn how to improve performance by
referring to the data but most adjustments will be incremental and explicit, easy to
understand and share.
Alternatively, predictive analytics leading to deeper insights through study of the
data is less likely in these industries. The reality is that little really unique, creative
knowledge or intelligence is discoverable in these old-line industries. Exceptions
always exist, but the track record shows few of those in recent years. Instead, the
impact of digital will be gradual, without large jumps. Artificial intelligence, if
employed, is likely to improve day-to-day decision-making but will have little impact
on larger, paradigm-changing strategic or tactical moves.
5 Conclusions
The digital economy, big data, and business analytics are all parts of a major change in
how firms will operate in coming years. The terms and concepts are sometimes fuzzy,
so organizations looking to deal with new modes of competition can understandably
feel uncertain as to their capabilities and how these might match up with new
opportunities. Reports such as the McKinsey piece better defining the impact of digital
developments, in defined stages, can be helpful to decision makers. We’ve taken those
concepts further, employing measures of the intangible assets of firms to better assess
how ready these organizations might be to participate in these digital stages.
These intangible assets have been viewed through different lenses but their basic
definitions are well-understood, and we do have some established metrics for judging
them. These metrics allow a firm to judge the typical level of the intangibles present in
industry competitors, their relative level of success in exploiting these intangibles, and
its own standing on this basis in the industry. In this paper, we’ve looked at intangible
assets categorized as data, explicit knowledge, tacit knowledge, and intelligence. Even
given some obvious exceptions in pharmaceuticals and, to a lesser extent, consumer
products, these heavy industries all show considerable data holdings but average to
below-average explicit knowledge and below-average tacit knowledge and intelligence.
The implications for managers in these industries are that the McKinsey report is
broadly right about participation in digital advances, including in the specific areas
noted such as efficiency of asset exploitation, and even worker-data interaction. But it’s
also important to note that most improvements from such applications will be more
incremental than paradigm-shifting. Day-to-day adjustments to processes due to data
monitoring is the likely path to success rather than heavy investment in the long-term
data/knowledge exploitation such as data mining or predictive analytics.
References
1. Polanyi, M.: The Tacit Dimension. Doubleday, New York (1967)
2. Nonaka, I., Takeuchi, H.: The Knowledge-Creating Company: How Japanese Companies
Create the Dynamics of Innovation. Oxford University Press, New York (1995)
3. Thomas, J.C., Kellogg, W.A., Erickson, T.: The knowledge management puzzle: human and
social factors in knowledge management. IBM Syst. J. 40(4), 863–884 (2001)
4. Matson, E., Patiath, P., Shavers, T.: Stimulating knowledge sharing: strengthening your
organizations’ internal knowledge market. Organ. Dyn. 32(3), 275–285 (2003)
5. Brown, J.S., Duguid, P.: Organizational learning and communities-of-practice: toward a
unified view of working, learning, and innovation. Organ. Sci. 2(1), 40–57 (1991)
6. Grant, R.M.: Toward a knowledge-based theory of the firm. Strateg. Manage. J. 17(Winter),
109–122 (1996)
7. Zack, M.H.: Developing a knowledge strategy. Calif. Manag. Rev. 41(3), 125–145 (1999)
8. Ackoff, R.: From data to wisdom. J. Appl. Syst. Anal. 16, 3–9 (1989)
9. Kurtz, C.F., Snowden, D.J.: The new dynamics of strategy: sensemaking in a complex-
complicated world. IBM Syst. J. 42(3), 462–483 (2003)
10. Simard, A.: Analytics in context: modeling in a regulatory environment. In: Rodriguez, E.,
Richards, G. (eds.) Proceedings of the International Conference on Analytics Driven
Solutions, pp. 82–92 (2014)
11. Rothberg, H.N., Erickson, G.S.: Big data systems: knowledge transfer or intelligence
insights. J. Knowl. Manage. 21(1), 92–112 (2017)
12. Manyika, J., Chui, M., Brown, B., Bughin, J., Dobbs, R., Roxburgh, C., Hung Byers, A.:
Big Data: The Next Frontier for Innovation, Competition and Productivity. McKinsey
Global Institute, New York (2011)
13. Sveiby, K-E.: Methods for measuring intangible assets (2010). http://www.sveiby.com/
articles/IntangibleMethods.htm
14. Tobin, J., Brainard, W.: Asset markets and the cost of capital. In: Nelson, R., Balassa, B.
(eds.) Economic Progress, Private Values, and Public Policy: Essays in Honor of William
Fellner, Amsterdam, North Holland, pp. 235–262 (1977)
15. Erickson, G.S., Rothberg, H.N.: Intangible dynamics: knowledge assets in context. In:
Hawamdeh, S., Chang, H.-C. (eds.) Analytics and Knowledge Management, pp. 325–354.
Auerbach/CRC Press, Boca Raton (2018)
16. Erickson, G.S., Rothberg, H.N.: Intelligence in Action: Strategically Managing Knowledge
Assets. Palgrave Macmillan, London (2012)
17. Noterdaeme, O., Schmitz, C., Sliczna, M., Somers, K., Van Niel, J.: Mapping heavy
industry’s digital-manufacturing opportunities. McKinsey & Co. (2018)
18. Chen, H., Chiang, R.H., Storey, V.C.: Business intelligence and analytics: from big data to
big impact. MIS Q. 36(4), 1165–1188 (2012)
19. McAfee, A., Brynjolfsson, E.: Big data: the management revolution. Harv. Bus. Rev. 90(10),
60–66 (2012)
Human-Organization Relation
in the Perspective of Industry 4.0
Waldemar Jędrzejczyk(&)
Faculty of Management, Czestochowa University of Technology,

Czestochowa, Poland
waldemar.jedrzejczyk@wz.pcz.pl
Abstract. The main aim of this paper is to present the prospective changes in
human-organization relation in enterprises caused by the technological progress,
the progressive global digital transformation of tools and processes in particular,
and to define the competences which should be currently developed in order to
maximize future competence matching. The paper indicates groundbreaking
innovations in the concept of Industry 4.0. It shows the most significant changes
in organizations caused by the fourth industrial revolution. The main focus was
the human-organization relation. Prospective changes in the functions of
employees and in the scope of competence needs of enterprises were determined
in the HR area. It was based on the literature analysis and the results of
empirical, primary and secondary research.
Keywords: Human Organization Industry 4.0 Relation Change
1 Introduction
Civilizational breakthroughs, globalization and the expansion of opportunities to

compete on the global marketplace, technical and technological progress, the devel-
opment of the knowledge-based economy as well as the uncertainty and high dynamics
of change that characterize these phenomena constitute economic realities in which
modern organizations have to function and cope with [1]. These phenomena determine
a great deal of changes in organizations. The most significant ones include changes of
organizational structures, models of operation and the working environment. This
translates into an increased complexity of management processes regardless of the kind
and type of the organization. However, it is not the end of changes, on the contrary.
Many researchers express a view that the greatest changes are still awaiting mankind.
E.g., Klaus Schwab, the founder and chairman of the World Economic Forum, believes
that “we are on the verge of a technological revolution, which will fundamentally
change the way we live, work and coexist. In its scale, scope and complexity this
transformation will be something humanity has not experienced yet…” [2]. These
changes in relation to production activity are referred to as the fourth industrial rev-
olution – Industry 4.0.
The term Industry 4.0 has become widely accepted in Europe. In some countries,
such as Poland, the term Industry 4.0 is used to describe a new conception of industrial
development. There are numerous synonyms for the Industry 4.0 conception in other

https://doi.org/10.1007/978-3-030-18715-6_2
Human-Organization Relation in the Perspective of Industry 4.0 15
countries, e.g.: Nouvelle France Industrielle – in France, Produktion 2030 – in Sweden,

Fabbrica Intelligente – in Italy, Made Different – in Belgium and the Netherlands,
Industria Conectada 4.0 – in Spain, Produktion der Zukunft – in Austria, Industry
Connected 4.0 – in the United States [3].
In the perspective of Industry 4.0 scientific studies increasingly often include
concepts such as: Management 4.0 [4], Human Resource Management 4.0 [5], Mar-
keting 4.0 [6], Digital Society 4.0 [7] and Economy 4.0 [8]. With regard to these terms,
the Industry 4.0 conception takes precedence, and the subsequent ones imply changes
in particular areas caused by technological progress.
The implementation of the Industry 4.0 conception in enterprises causes changes
both in manufacturing and management systems. The relation between human and
organization is also subject to transformation. Along with the development of
automation and robotization, people are “eliminated” from the direct process of
manufacturing, which will shape new relations between the organization and its human
resources. Behaviors and functions of employees, both managers and executive
workers, will change.
The main aim of this article is to present the prospective changes in the human-
organization relation in enterprises caused by the technological progress, the pro-
gressive global digital transformation of tools and processes in particular, and to define
the competences which should be currently developed in order to maximize future
competence matching.
2 The Prospect of the Fourth Industrial Revolution

in Enterprises
The following transformations such as: “mass” digitization of processes, decision-

making based on virtual simulations and real-time data processing, human-to-machine
and machine-to-machine communication and new manufacturing technologies are
referred to as the fourth industrial revolution on the basis of which the Industry 4.0
conception was developed [9, 10]. The current “transformation” is preceded by prior
technological changes related to mechanization of manufacturing processes – industry
1.0, electrification – industry 2.0 and computerization – industry 3.0. The Industry 4.0
conception was initiated in Germany in 2011–12 [11, 12]. It is based on the assumption
that internal processes in enterprises should be digitized, automated and robotized to
the greatest extent in order to make them faster and more efficient than hitherto [13].
Enterprises are to evolve towards smart factories, which is expected to result in a sharp
surge in their productivity [14]. It will consequently lead to an intermingling of the real
and virtual world [15, 16]. The number of companies in the world with a high level of
digital development, which is characterized by the ability to combine current activities
from the virtual word with the real world, is increasing. According to the research
conducted by Zebra Technologies Corporation – a manufacturer of technical printing
equipment, including devices for printing and programming RFID radio tags – the
number of companies defined as “The Intelligent Enterprises” doubled to 10% in 2018
[17]. Breakthrough innovations in the Industry 4.0 conception include: machine
learning, predictive analysis, blockchain, virtual and augmented reality as well as the
16 W. Jędrzejczyk
Internet of Things and Internet of Services, which along with Big Data will create a
new brand of artificial intelligence – cyber-physical systems (CPS)1 [18, 19]. These
technologies are being increasingly used in the organizational practice [20], and until
recently they have been presented as visions for the distant future. Many technologies,
which will certainly be the main driver of innovativeness and can change the direction
of business and social policy development, are still in the phase of research or first
market tests. For example, researchers from the Massachusetts Institute of Technology
classified 10 breakthrough technologies in 2018, including: 3D metal printing, artificial
embryos, sensing cities, artificial intelligence for everybody, dueling neural networks,
speech and context recognition in real time expression, perfect online privacy,
reduction of CO2 emission in electrical energy production, disease risk analysis of
genetic material and materials’ quantum leap [21].
It can be assumed that comprehensive digitization and optimization of internal
processes constitute one of the stages of exploiting the potential that new technologies
bring. It contributes to the creation of innovative products and services and new
business models [22], which can certainly have a much greater impact on the level of
competitiveness of enterprises than process optimization alone [23]. It is necessary to
involve all the company employees in the process of creating innovative products, not
just the team responsible for innovations.
The fourth industrial revolution will result in a paradigm shift from the “centralized
production” to “decentralized production”. In the paradigm of the “centralized pro-
duction”, complete production tasks are conducted in a single enterprise. Such systems
are based on centralized management and control. The paradigm of decentralization
means a departure from centralized management and control [24]. “Networks of
autonomous production resources capable of controlling themselves in responding to
different situations, self-configuring, knowledge-based, sensor-based, spatially dis-
persed and including appropriate planning and management systems” will arise [13,
25]. Functioning in the network imposes on enterprises a need to focus on their own
key competences and to shift other activities to their business partners [24].
The human-machine relation will also change. The development of intelligent
interfaces will transfer into a higher level of “cooperation” between man and
machine/robot. The function of the employee will evolve towards “managing” the
activities performed by the cooperating machines. People will become so-called
“digital conductors” [23]. The human-machine system will be an integrator of con-
textual human decision making and the precision and regularity of machines.
The challenges that are already emerging may not be too unambiguous and clear
yet, but even such challenges indicate that the transformation processes associated with
the implementation of the Industry 4.0 idea will be of a revolutionary rather than
merely evolutionary nature [7].
Numerous market analysts treat digitalization as a new factor in development of
countries [26, 27]. There are also those who do not treat the digitization phenomenon in
1
A cyber-physical system is a combination of hardware, software and communication elements, with
a particular emphasis on communication. The idea is for machines to become more intelligent: they
would communicate with each other, use shared data and create their own “social networks”.
the category of opportunities, but in the category of “to be or not to be”. Compared
with other Central and Eastern European countries, Poland has worse indicators, e.g. in
terms of digital skills of the society in all age groups and the use of Big Data tools or
cloud computing by enterprises from the SME sector. Poland is struggling with a
shortage of breakthrough technologies. Most initiatives have a negligible importance
for the economy – they are only micro-innovative. That is why the Polish government,
similarly to other EU countries, has established a new entity in Poland to act as an
integrator – the Future Industry Platform Foundation, whose task will be to intensively
disseminate knowledge and skills necessary to function in the realities of Industry 4.02.
3 Changes in Human-Organization Relation in Enterprises

in Perspective of the Fourth Industrial Revolution
The Industry 4.0 conception determines significant changes in the relation between
man and organization. New business models will require a re-examination of compe-
tence models and corporate culture of enterprises in the light of expected new com-
petences and the need to attract and maintain adequate human capital [2]. The ongoing
transformation of the organization triggers the need for significant changes, especially
in organizational structures, mechanisms of leadership, corporate culture and compe-
tences, both organizational and individual [28]. The modern model of workplace
presence will be changed due to the ability to remotely communicate with devices and
to control their work. This will provide workers with an opportunity to flexibly con-
figure their physical presence time at the workplace. It will lead to a creation of new
business models in the field of participation in online production processes.
The activities of organizations are currently focused on the aspiration to increase
the effectiveness of people in the organization. As a result, they mainly focus on
developing proper methods of motivation, control and human resource management in
the broad sense. In the situation when optimization of organization’s operations results
from the digitization of processes (automation, computerization and robotization),
employee’s functions in the organization will change. It will entail the need to possess a
greater deal of different competences than those needed today. The function of pro-
fessional competences will decrease in favor of mental and interdisciplinary compe-
tences such as: entrepreneurship, creativity, communicativeness and teamwork. The
abilities to communicate, both in the human-to-human (P2P) and human-to-machine
(P2M) relation, to search for knowledge, its creative and innovative creation, will
become more significant than the abilities to pass or use the possessed knowledge.
As it has been already indicated in the introduction to the present paper, new
technical solutions will lead to the “elimination” of people from the direct process of
manufacturing, which will result in the need to redefine the profiles of existing
workstations, create completely new jobs and possess new competences such as skills
enabling the “operation” of intelligent machines/systems and the skills necessary to
2
The Future Industry Platform Foundation bill was adopted by the Polish government in October the
23rd 2018.
18 W. Jędrzejczyk
service and repair them. Technological development creates a demand for new occu-
pations, for which technologies become work tools. Competences development will be
perpetual. It will be integrated into production processes. It is assumed that it will be
supported by the introduction of digital employee assistants.
3.1 Change in the Function of the Employee in the Enterprise

Creating new professions and positions or redefining existing ones and assigning the
desired competences to them will require a look at the organization, its every branch,
employee and task from the perspective of the “human domain” and “machine
domain”. When creating a new working environment it is necessary to define the
activities which will be performed by people and the ones which will be performed by
machines in order to assign the roles properly. Therefore, it is necessary to analyze the
strengths of people and machines in the first place and then to divide the tasks so as to
highlight the elements that can be done better by man and the ones that can be handled
better by machines. The implementation of this task requires a creative and imaginative
approach. Do not be afraid to assign to machines all the tasks that they can do better
than a human being.
The elementary human domains include above all: entrepreneurship, innovative
and creative thinking, heuristic reasoning, intuition, generating innovations and
empathy. The elementary machine domains include the following features:
• they can easily perform tedious, repetitive tasks – the machine performs repetitive
tasks much more precisely than a human being and does not make mistakes,
• they are faster and more efficient in mathematical, logical and statistical operations
than human,
• they can easily perform tasks that require rapid analysis (in real time) of large
amounts of data coming from many different sources,
• they analyze the answers to structured questions more quickly and are “able” to
answer them.
New technologies are used not only in the processes of direct manufacturing of
products, but also in many other areas of the enterprise’s operations such as customer
service or the provision of technical support services – relieving consultants in repetitive
activities. Chatbots, computer programs which simulate “living users” operations, are
increasingly being used in these areas. They replace humans with simple and repeatable
conversation patterns3. Their precision and practicality will surely increase over time.
It does not mean there are no barriers to the development of this trend. The biggest
barrier is people themselves, the real human nature. According to the research con-
ducted by Capgemini, consumers prefer interactions with humans, which can be
3
For example, the Polish company Budmimex, which employs almost 5,000 employees, intends to
use a chatbot in the recruitment process, with which candidates will be able to communicate via a
facebook Messenger. A digital recruiter will ask every applicant about their skills, knowledge of
foreign languages and financial expectations. Responses will be compared with the employer’s
requirements and capabilities – an objective assessment of candidates. Those of whom the robot
accepts, will receive invitations for interviews in the real world.
supported by artificial intelligence tools, rather than with machines themselves – 55%
of those surveyed in Europe and the USA [29].
New technologies will not be able to replace a human being in all areas and aspects
of the organization for a long time4. They have not yet reached a sufficient level of
maturity to compete with human creativity, flexibility and intuition. Advanced ana-
lytical tools and artificial intelligence can only support human knowledge and cre-
ativity. The inalienable domain of people will continue to be, among others, creation of
new business models and innovations – innovative products and services.
The Industry 4.0 conception in relation to the HR area of the organization should be
properly perceived. The following thesis can be boldly proposed: new technical
solutions are aimed at increasing the capabilities of employees and not at replacing
them [30]. A replacement of man in manufacturing processes ought to be perceived not
as a threat to humanity but as an opportunity. Thus, man in the organization will be
able to primarily focus on what he does best – what is his domain.
3.2 Competence Needs of Enterprises

“The demand” for new competences (finding employees with the desired competences)
as a result of technical and technological progress will be one of the most significant
problems in the HR area, which future organizations will have to deal with. “The
demand” for key interdisciplinary competences in organizations is already high,
whereas there is a “deficit” of them in the labor market. For example, entrepreneurs
highly evaluate the usefulness of such competences as: entrepreneurship, creativity,
communicativeness and teamwork – the average level of usefulness of the highlighted
competences is 4.2 on a scale from 1 – competence completely not needed, to 5 –
competence needed very much and systematically used (Fig. 1). On the other hand, the
level of their absorption by people is relatively low – the average level of mastery of the
highlighted competences in the surveyed group of students is 3.2 on a scale from 0 –
competence not mastered to 5 – competence mastered at very high level (Fig. 2)5.
4
For the time being, the horizon 2030 is indicated. Until then, abilities such as: intuition, thinking and
emotional intelligence are characteristic only for people. However, it cannot be excepted that in the
long term, machines will also be characterized by them.
5
Research conducted within the framework of a scientific project “The acceleration method of
development of transversal competences in the students’ practical training process” – http://www.
awt.org.pl/projekty/o-projekcie/?lang=en. The project leader was Poznan University of Technology –
a higher education institution, and the partners include: Czestochowa University of Technology
(represented, among others, by the Authors of the present paper), Wroclaw University of Economy,
Western Chamber of Industry and Commerce, in Gorzow Wielkopolski, Centria - ammattiko-
rkeakoulu (Finland), JEDU Jokilaaksojen koulutuskuntayhtymä (Finland), Univerza v Mariboru
(Slovenia) and Univerzita Mateja Bela v Banskej Bystrici (Slovakia). The project was conducted
within the Erasmus+ programme – the key action: Cooperation for innovation and good practice.
Project implementation time: from October the 1st 2015 to August the 31st 2018. The main aim of the
project was to develop and implement an innovative method accelerating the development of stu-
dents' transversal competences by improving the practical education use. In the project the analyses
were narrowed down to four transversal competences (the effect of the adopted project assumptions):
entrepreneurship, creativity, communicativeness and teamwork. The author of this paper was a
research team member.
20 W. Jędrzejczyk
Fig. 1. Competence suitability assessment – aggregated results obtained in individual countries.

Source: own elaboration.
Fig. 2. The level of possessed competences – aggregated results obtained in individual

countries. Source: own elaboration.
Therefore, it is not easy to find proper employees anymore. According to the

ManpowerGroup studies [31]:
• 52% of Polish companies declare that they have problems with finding “talented
people”,
• 8% - report deficiencies in hard skills,
• 1% - report deficiencies in soft skills and communication skills,
• 12% - say that the candidates lack experience,
• 18% - say that there are employees who meet all the requirements of HR directors,
but they value themselves too high.
ManpowerGroup’s results are based on Polish realities, but it is highly probable
that the global economy is facing a similar problem.
The competence gap may relate to different skill categories. A very important
category of competence are competences in the field of new technologies. Surveys
conducted by PricewaterhouseCoopers show that enterprises see a growing competence
gap in the area of technical skills related to new technologies. Respondents in the
survey indicated that their companies lacked suitably qualified teams (63% of
respondents), most notably in the areas of cyber-security as well as data protection and
data analytics. They also predicted that this gap would widen in the future (39% of
respondents) [32]. Reducing the competence gap in this area will also be one of the
priority challenges facing enterprises. Management should be aware of the fact that the
digital transformation is a complex and costly process that requires a lot of resources.
Only those companies that are able to use technology in a creative way will achieve
success.
The paper shows that the competences which will be desired in new human-
organizational relations – as a result of technological progress referred to as the fourth
industrial revolution – are primarily mental, communication, interpersonal and lead-
ership competences. They can be used for various types of tasks in many areas of
company’s operation. Such competences are also referred to as transversal, interdis-
ciplinary competences. These competences are also desirable in the current economic
reality, which is confirmed by the results of the own research as well as the results of
research by other authors/institutions, whereas there is a “deficit” of them in the labor
market. The problem will escalate. Therefore, it is extremely important to define the
“future” competences already today and to develop them at the various stages of
education. Such actions will prevent the competence gap from widening.
4 Discussion
It seems that the most appropriate strategy for modern organizations is ambidexterity
strategy [33]. It consists in balancing exploitation activities (short-term perspective,
current tasks orientation) and exploration activities (long-term perspective, future ori-
entation, search for opportunities and threats in the long time span), seemingly con-
tradictory, in order to obtain benefits from both. The ambidextrous approach to the
strategy cognitively corresponds to the phenomenon which have been defined as a
dualism in the organization management by theoreticians. On the one hand, organi-
zations are forced to run their current business perfectly, but on the other hand, to
manage better strategic changes. In this situation, the organization management and
change should be concurrent and not sequential [34].
Exploratory activities are considerably more difficult than exploitation activities.
According to the uncertainty management concept [35], various scenarios for the
unknown future (from more realistic, through less realistic, to almost impossible),
which are the trigger of the built or modified organization’s strategy, including HR,
should be taken into account. The organization’s competence potential analysis should
focus not only on the current competence needs, but also on the prospective ones,
which may turn out to be crucial in the future.
Currently, the most realistic scenario for the new technologies development is the
state in which machines, devices and products (i.e. things) will communicate with each
22 W. Jędrzejczyk
other in order to cooperate [36], without human intervention. Therefore, major efforts
are being made in this direction, both in the science and business field. Intensive design
and experimental work is being carried out, particularly in the science field, in order to
implement given concept. It is one thing to develop an idea and another to implement it
[37]. According to this scenario, the direction of employee competences development
is also postulated – both those that enable the new solutions creation (conceptual,
technical and engineering competences) and those that will be essential due to changes
in the people’s fundamental roles in organizations, taking place under their influence
(conceptual, technical and interdisciplinary competences). It is certain that both hard
and soft competences will be desirable in these actions – apart from professional and
engineering competences, such skills as intuition and creativity are also necessary,
which in combination with innovativeness will turn into innovation.
Given scenario presents an assumption that “things” will communicate with each
other in “good faith”, i.e. in order to optimize the production or service processes. But
is this the only possible scenario? It cannot be excepted that in the future “things” will
also communicate with a destructive attitude in order to harm other entities, bearing in
mind their “own” benefit above all. An even worse variant is possible – intelligent
machines will want to take control of people and the world. Such a scenario, although
unlikely to happen, should also be taken into consideration.
In order for the first scenario to come true, the future competences need to be
shaped appropriately. People need to be prepared to anticipate. They should focus even
more on the intuition and other characteristics that constitute human nature. They must
also acquire knowledge and skills in new technologies. Competences should be shaped
in such a way that the domain of the Industry 4.0 concept would be better cooperation
between man and technology, not competition with technology for jobs.
Such competence building requires systemic actions in which the state and
supranational organizations play a significant role. It is desirable, for example, to
integrate graduates of technical and economic universities, i.e. to integrate engineering
and scientific competencies with managerial and interdisciplinary competencies.
5 Summary
The use of new technologies in the Industry 4.0 conception will determine changes in
practically every area of enterprise activity. The impact of technological factors is
expected to be stronger than the influence of economic and market factors. Techno-
logical factors influence not only the organization of production processes and prod-
ucts, but also relations with customers, partners in supply chains and competitors. As a
result, enterprises are forced to develop new strategies and new business models. New
technologies also have an impact on human resources management, organizational
structure of the enterprise and work organization. As technology advances, the working
environment evolves, which requires employees to possess new and specific
competences.
The digital transformation in the HR area brings both opportunities and challenges
for organizations and employees. New technologies and automation will not replace
human competencies, but they force employees to be more flexible. The ability to
retrain and acquire new skills in order to keep pace with the changes driven by the
digital transition will be a key factor in staying in the labor market. The biggest threat
to the business sector will be the lack of people with the right set of competences. The
talent management strategy in organizations needs to change to respond better to the
challenges of the future.
The paper is a synthetic presentation of theoretical and empirical analyses in the
scope of the Industry 4.0 conception and its prospective impact on the functioning of
enterprises, concerning the formation of human-organization relations in particular.
The paper poses a significant contribution to further research and empirical analyses in
this area, as it is difficult to predict exactly what further changes will appear in the
incoming years. There are also no proven solutions in management theory that show
how to prepare both organizations and employees for the changes ahead.
References
1. Ju, K.-J., Park, B., Kim, T.: Causal relationship between supply chain dynamic capabilities,
technological innovation, and operational performance. Manage. Prod. Eng. Rev. 7(4), 6–15
(2016)
2. Schwab K., The Fourth Industrial Revolution. Crown Business, New York (2017). https://
www.amazon.com/Fourth-Industrial-Revolution-Klaus-Schwab/dp/1524758868. Accessed
26 Nov 2018
3. Ślusarczyk, B.: Industry 4.0: are we ready? Pol. J. Manage. Stud. 17(1), 232–248 (2018)
4. Oswald, A., Muller, W. (eds.): Management 4.0 – Handbook for Agile Practices. Books on
Demand, Norderstedt (2017)
5. Bruch, H., Färber, J., Block, C.H.: HRM 4.0. Personalarbeit dr Zukunft. TOP JOB-
Trendstudie 2017, Universität St. Gallen (2017)
6. Kotler, P., Kartajaya, H., Setiawan, I.: Marketing 4.0: Moving from Traditional to Digital.
Wiley, Hoboken (2016)
7. Kiepas, A.: Man and an intelligent organisation in the perspective of industry 4.0 (in Polish).
In: Kiełtyka, L., Jędrzejczyk, W., Kobis, P. (eds.) Modern Management Challenges. New
technologies – Innovativeness – Competences, Toruń, 11–24 (Dom Organizatora) (2018)
8. Economy 4.0. Die Digitalisierung der Wirtschaft. In Pact Mediaverlag, November 2017
9. Lasi, H., Fettke, P., Kemper, H.G., Feld, T., Hoffmann, M.: Industry 4.0. Bus. Inf. Syst. Eng.
6(4), 239–242 (2014)
10. Bartodziej, C.H.J.: The Concept Industry 4.0: An Empirical Analysis of Technologies and
Applications in Production Logistics. Gabler Verlag, Wiesbaden (2017)
11. Pereira, A.C., Romero, F.: A Review of the meanings and the implications of the industry
4.0 concept. Proc. Manuf. 13, 1206–1214 (2017)
12. Hariharasudan, A., Kot, S.: A Scoping review on digital english and education 4.0 for
industry 4.0. Soc. Sci. 7(11), 227 (2018)
13. Zawadzki, P., Żywicki, K.: Smart product design and production control for effective mass
customization in the industry 4.0 concept. Manage. Prod. Eng. Rev. 7(3), 105–112 (2016)
14. Zhong, R.Y., Xu, X., Klotz, E., Newman, S.T.: Intelligent manufacturing in the context of
industry 4.0: a review. Engineering 3(5), 616–630 (2017)
15. Häckel, B., Hänsch, F., Hertel, M., Übelhör, J.: Assessing IT availability risks in smart
factory networks. Bus. Res. 1–36 (2018). Open Access
24 W. Jędrzejczyk
16. Kobis, P.: Virtualization of information resources in organizations: the era of cloud
computing (in Polish). Przegląd Organizacji 2(901), 34–42 (2015)
17. Zebra technologies study: the number of intelligent enterprises has doubled since 2017 (in
Polish). https://www.zebra.com/pl/pl/about-zebra/newsroom/press-releases/2018/zebra-stud
y-finds-number-of-intelligent-enterprises-doubles-since-2017.html. Accessed 28 Nov 2018
18. Colombo, A., Bangemann, T., Karnouskos, S., Delsing, J., Stluka, P., Harrison, R., Jammes,
F., Lastra, J.L. (eds.): Industrial Cloud-Based Cyber-Physical Systems: The IMC-AESOP
Approach. Springer, Switzerland (2014)
19. Lee, J., Bagheri, B., Kao, H.A.: A cyber-physical systems architecture for industry 4.0-based
manufacturing systems. Manuf. Lett. 3, 18–23 (2015)
20. Olszak, C.M.: Creation of big data-based business value (in Polish). Przegląd Organizacji 7
(942), 35–45 (2018)
21. Breakthrough technologies (2018). https://www.technologyreview.com/lists/technologies/
2018/. Accessed 28 Nov 2018
22. Kenney, M., Rouvinen, P., Zysman, J.: The digital disruption and its societal impacts. J. Ind.
Compet. Trade 15(1), 1–4 (2015)
23. Ustundag, A., Cevikcan, E.: Industry 4.0: Managing The Digital Transformation. Springer,
Switzerland (2018)
24. Kelly, R.: Constructing Leadership 4.0: Swarm Leadership and the Fourth Industrial
Revolution. Springer (2019). Palgrave Macmillan
25. Kagermann, H., Helbig, J., Hellinger, A., Wahlster, W.: Recommendations for implementing
the strategic initiative INDUSTRIE 4.0: securing the future of german manufacturing
industry. Final Report of the Industrie 4.0 Working Group, Forschungsunion (2013)
26. Basl, J.: Pilot study of readiness of czech companies to implement the principles of industry
4.0. Manage. Prod. Eng. Rev. 8(2), 3–8 (2017)
27. Nagy, H.: A role for the state in the digital age. J. Innov. Entrep. 7(5), 1–16 (2018)
28. Devezas, T., Leitão, J., Sarygulov, A. (eds.): Industry 4.0: Entrepreneurship and Structural
Change in the New Digital Landscape. Springer, Heidelberg (2017)
29. The Secret to Winning Customers’ Hearts With Artificial Intelligence. Add Human
Intelligence, Capgemini Digital Transformation Institute. https://www.capgemini.com/de-de/
wp-content/uploads/sites/5/2018/07/Survey-Report-AI-in-Customer-Experience.pdf. Acces-
sed 30 Nov 2018
30. The Next Era of Human-Machine Partnerships. http://www.iftf.org/humanmachinepartne
rships/. Accessed 26 Nov 2018
31. 2018 Talent shortage survey, manpowergroup, solving the talent shortage
32. 2017 Global Digital IQ® Survey. https://www.pwc.pl/pl/pdf/publikacje/digitaliq2017-
poland-key-findings.pdf. Accessed 30 Nov 2018
33. Zakrzewska-Bielawska A.: Enterprises development strategies: a new look (in Polish). PWE,
Warszawa, (2018)
34. Abell, D.F.: Competing today while preparing for tomorrow. Sloan Manag. Rev. 40(3), 73–
81 (1999)
35. Collins, J., Hansen, M.T.: Great by Choice: Uncertainty, Chaos, and Luck – Why Some
Thrive Despite Them All. Harper Business, New York (2011)
36. Kiełtyka, L., Zygoń, O.: Contemporary communication forms of everything with everyone –
how to manage with the use of the internet of things and internet of everything (in Polish).
Przegląd Organizacji 2(937), 24–33 (2018)
37. Niedbał, R., Wrzalik, A., Sokołowski, A.: Fourth industrial revolution as a challenge to
maintain enterprise competitiveness (in Polish). Mark. i Rynek 7, 557–570 (2017)
Edge Computing Technologies as a Crucial
Factor of Successful Industry 4.0 Growth.
The Case of Live Video Data Streaming
Joanna Gąbka(&)
Department of Laser Technologies, Automation and Production Engineering,

Wroclaw University of Science and Technology,
27 Wybrzeże Wyspiańskiego Street, 50-370 Wrocław, Poland
joanna.gabka@pwr.edu.pl
Abstract. This article presents edge computing as a new approach which can
significantly benefit development of Industry 4.0 in different areas. The edge
computing was explained in comparison with cloud computing. Its main
advantages can be obtained implementing both edge and cloud computing as a
hybrid solution. The advantages of the new data management method were
assigned to the topics comprised within each of the three differentiated main
research areas built in the Industry 4.0 paradigm: Individualization of produc-
tion, Horizontal integration in collaborative networks, End-to-End digital inte-
gration. The tables designed show impact of the advances resulting from edge
computing on the development possibilities in manufacturing and production
while realizing the most probable exploitation and growth scenarios. To make
the predictions based on the literature more realistic there was presented a case
study of the project where edge device for stereoscopic 180° live video
streaming was planned to be implemented bringing significant improvements
into the data capturing and management process. The algorithmic possibilities
presented may be analogically exploited in the topics mentioned in all the three
main Industry 4.0 theme groups.
Keywords: Edge computing Industry 4.0

Live stereoscopic video data streaming
1 Introduction
Industry 4.0 is described as a next important step in the evolution of production and all
the connected fields. It is very complex movement consisting of many fields. In the [1]
authors indicate three main research streams supporting the development process with
numerous topics contained in each of them. The mentioned taxonomy is as following:
1. Individualization of production which is closely correlated with such areas as mass
customization, modularization, Flexible Manufacturing Systems (FMS) and Recon-
figurable Manufacturing Systems (RMS), Distributed control, Self-optimization, Rapid

https://doi.org/10.1007/978-3-030-18715-6_3
26 J. Gąbka
Manufacturing, Cloud computing; 2. Horizontal integration in collaborative networks

comprising Collaborative Networks, Distributed Manufacturing, Supply – Chain
Flexibility, Supply – Chain Visibility, Internet of Things and Services; 3. End-to-End
digital integration correlated with Virtualization of the process chain, Individualized
Traced Data, Real-Time Operating Systems, Simulation and Modeling of products and
processes, Simultaneous planning of products and production processes, Value Added
Services. If we take closer insight into each of the topics it will be obvious that fast
collecting, transferring and processing huge quantities of data from different sources is
the essential condition of progress in any of the specific research fields. The cloud
computing indicated within the first research stream actually became visible in all other
places [2]. Its importance is especially highlighted when Internet of Things (IoT) is
discussed [3, 4]. The practical implementation of the Cloud computing altogether with
IoT shown limitations connected with bandwidth, costs, speed, predictive analytic,
remoteness and maintenance [5–7]. As an answer to this drawbacks and edge com-
puting approach was brought into being. It is based on idea of devices which inter-
mediate layer between the equipment used and the cloud database [8, 9]. This kind of
solutions indicate great potential and are a chance for acceleration of the Industry 4.0
growth. This article presents technologies connected with edge computing, provides
definitions, shows a case of study of the particular device usage for video data
streaming and potential of such solutions for manufacturing.
2 Cloud Computing vs. Edge Computing
Cloud computing is defined by The National Institute of Standards and Technology as

“model for enabling ubiquitous, convenient, on-demand network access to a shared
pool of configurable computing resources (e.g. networks, servers, storage, applications
and services) that can be rapidly provisioned and released with minimal management
effort or service provider interaction” [10]. Due to the rapid increase in demand for data
storage and processing this solution became popular in different domains and spe-
cializations both in industry and private sector [11]. It gained its position offering low
price, easy management and the transparency of clearance resulting from pay-for-use
strategy, distributed and uninterrupted services, delivering fast unlimited space and
powerful computing capacity for data storage and administering [12]. In addition it is
associated with green computing because of economic energy, computing space and
power use. It is also associated with effective remedy to the crises providing reliable
backup. There are three main service models referring to the cloud computing [13]:
1. Software as a Service (SaaS) – in this model cloud offers software, service or
application which can be used after subscription. The delivery is made directly
through network on-demand. The most popular SaaS in this category are Google
Docs and Spreadsheets and IBM LotusLive.
2. Platform as a service (PaaS) – it provides set of runtime environments hosted on the
provider’s server. After paying certain price the platform and framework for soft-
ware deployment is given to application developers, implementers, testers and
Edge Computing Technologies as a Crucial Factor of Successful Industry 27
administrators which shows that this model operation covers all phases of software
lifecycle. Here the well-known commercial are Amazon AWS and Microsoft Azure.
3. Infrastructure as a service (IaaS) – is a collection of servers, storage spaces and
networks. This model is very closely correlated with virtualization which was
mentioned among the main topics included in Industry 4.0. In this model Cloud
Service Provider (CSP) gives an access to storage, network, routers, servers through
virtual desktop. The transaction value is calculated on the base of utilization in
different dimensions such as Central Processing Unit (CPU) hour, data GB stored
per hour, additional services like scaling, network bandwidth consumed, network
infrastructure per hour. The exemplary services available on the market are Amazon
S3 and EC2 and VMWare.
In each of these three categories user can find such qualities as elasticity which is
appreciated due to dynamic working conditions in today’s environment and resource
pooling helping risk management.
The service models are distributed in four different ways: public/external cloud
available to all users. It offers services to the third part acquiring spare capacities from
various enterprises such as Sun Cloud, Windows Azure Service Platform and others.
The second way preferable in industry due to the security issues is private/internal
cloud build specially for the particular organization which is exclusively authorized to
control and owns the solution. Another cloud deployment form is hybrid cloud/virtual
private cloud (VPC) with combination of both previous approaches. Some system is
hosted and the rest of it managed externally. The last way of the cloud distribution is
community cloud where both on the user and management side can be many related
organizations. The service models and distribution options are presented in the Fig. 1.
The cloud computing has also some disadvantages and implemented risks, mostly
connected with security, such as data loss, data cleaning, account hijacking, limited
control over the process, unwanted data inspection by CSPs, lack of portability/
migration from one service provider to the other, lack of auditability, issues resulting
from diversified quality of service [14, 15]. One of the important flaws in context of
industry is long response time resulting from raw data transferring, processing and then
sending back to the device connected. The fast growth of demand for such a services
resulting also from the digitalization of Industry led to the significant challenges with
virtualization, interoperability, resource scheduling, multi-tenancy, load balancing and
most importantly security [16, 17].
This issues altogether with fast development of this field motivated to create a new
solution. As an answer to the identified shortcomings a fast evolution of the cloud
computing occurred towards solutions known as edge computing. The other termi-
nology associated with this topic contains such a terms as fog, mist computing and
cloudlets.
28 J. Gąbka
SoŌware as a Service Plaƞorm as a Service Infrastructure as a

Community
(SaaS) (PaaS) Service (IaaS)

Cloud
CustomizaƟon
Public Cloud
CustomizaƟon
CustomizaƟon
Hybrid Cloud
ApplicaƟon
ApplicaƟon
Public Cloud
Plaƞorm
Fig. 1. Combination of cloud computing service models and distribution options.
Edge computing is based on the idea of processing data on the dedicated device
nodes known also as smart devices or edge devices which are placed on the edge of
core process enabling fast calculation which leads to the particular action execution
before the communication with a cloud is performed. The edge computing is rather
enrichment and modification to the cloud computing paradigm rather than its
replacement because it may also use the cloud to the less sensitive data collecting or
less critical decision making. The edge devices are more in charge of the decision
making in cases when instant response is required. The division of computing and
storage tasks between edge device and cloud is dependent on conditions in which their
service. Figure 2 shows the main difference between this two approaches.
It is easy to notice in the Fig. 2 that the main difference is additional element of the
architecture on the side of edge computing. This small modification solves many
problems identified within cloud computing and implies new possibilities like more
autonomous factories able to operate more effectively without human involvement. The
quick on spot computations enable to deal with problems such as downtimes in the
more robust way. The less sensitive data can still be operated by traditional cloud
computing especially in cases when there is a will to wide access to them from
dispersed locations. It is likely that edge computing will complement cloud computing
instead of replacing it. International Data Corporation research forecast shows that 45%
of data produced by IoT devices will be stored, processed and analyzed on the edge of
networks in the next three years. This gives over 6 billion appliances connected to the
edge computing devices giving quick feedback and impacting the execution level.
Cloud CompuƟng Edge CompuƟng
Cloud server Cloud server

ComputaƟon
Reasoning
Data Storage
Internet Internet
Internal databasis Internal databasis
Crucial computaƟon
Quick response requiring
reasoning
SensiƟve data storage Edge device
Devices/Systems/IoT serviced
Fig. 2. Cloud computing in comparison with edge computing.
3 Edge Computing as an Important Element of Industry 4.0

Development
The findings regard first themes group – Individualization of production are presented
in Fig. 3. The edge computing will bring more clarity towards data safety. Sensitive
information from the customer may be managed and protected which can result in
higher amount of clients willing to share date and buy customized products. On the side
of manufacturer it means more timing in delivery due to the local fast decision making
and more flexibility resulting from independence of outside service provider. The
development of modularization may benefit thank to the integration of the design
function and manufacturing process. The information gathered from previous series of
products may be used to modify construction of the next series. FMS&RMS is the topic
that seems to benefit most from edge computing. The quick communication between
machines gives more reliability and independence of the system from human work-
force. The problematic field of rescheduling may be solved more efficiently. The
production system will be less prone to the attacks from outside, the human inter-
vention will be minimized. All of this gives opportunity for more internal system
integration and autonomous workload balancing which in turn leads to the cost
30 J. Gąbka
Industry 4.0 Fast response Robust data Filtering sen- Processing lo-
connected time management sitive data locally saving
topics (minimizing cally computational
data loss and power and
operational bandwidth
failure)
edge
computing
advantages
Mass cus- Customized More trust due Cost effective-
tomization services gener- to personal da- ness less relia-
ated faster ta safety bility on out-
(crucial in sourced service
healthcare)
Modulari- Cohesive de-
zation sign/productio
n processes
FMS & Efficient Production Less need of Huge advance
RMS communication systems less human inter- in production
between ma- vulnerable to vention into system integra-
chines; hackers attacks the system tion and capac-
Individualization of production
Quicker reac- ity load balanc-

tion to the in- ing
stant changes
in the produc-
tion schedule
Distributed More precise Historical data and accurate
control data analysis gathered local- calculations in
due to the lack ly always the necessary
of delay available field only
Self- Efficient load More possibili-
optimiza- balancing ties in re-
tion sources capaci-
ty utilization
Rapid Advancements Correcting
Manufac- in work on dig- process param-
turing ital twin and eters online
security issues
connected
Cloud Less pressure Less responsi- Less security Cost savings
computing on responsive- bility on the concerns
ness from cus- side of CSP
tomers
Fig. 3. Cloud computing advantages as growth factor in individualization of production.

failure)
edge
computing
advantages
Collabora- More flexibil- More integrity Mutual trust Cost effective-
tive Net- ity and time ef- internal and due to sensitive ness less relia-
works ficiency in co- external data internal bility on out-
operation protection sourced
services
Less require-
Horizontal integration in collaborative networks
ment for man-

ual data input
Distributed Less human in- Easy access to Mutual trust More transpar-
Manufac- teraction in data, undis- due to sensitive ent rules to-
turing checking turbed, quick data internal wards opera-
available ca- data exchange protection tional cost of
pacities collaboration.
Supply – Fast modifica- Increased reli- Less need of Cost effective-
Chain Flex- tion in re- ability, in- human inter- ness
ibility sponse to the creased effi- vention into
changes ciency the system
Supply – More trust due Historical data More control

Chain Visi- to the reliable gathered local- over data (ac-
bility answers and ly always cess and au-
subsequent ac- available for thorization eas-
curate actions audits or pro- ier to be done)
cedure im-
provements
Internet of Smarter cities More control Easy populari- Cheaper solu-
Things and (securi- over data zation of the tions common-
Services ty/communicat More reliable IoT thanks to ly used by so-
ion/services) actions the information ciety.
Smarter houses security.
More possibili-
(safe- ties for AI
ty/comfort) (self-learning
Smarter algorithms)
healthcare sys-
tems and de-
vices (integrat-
ed/balanced
approach, early
warnings be-
fore emergen-
cy)
Fig. 4. Cloud computing advantages as growth factor in Horizontal integration in collaborative

networks.
32 J. Gąbka
failure)
edge
computing
advantages
Virtualiza- More embed- Less failures Increased secu- Cost effective-
tion of the ded intelli- due to commu- rity, increased ness less relia-
process gence and au- nicational dis- trust bility on out-
chain tonomous turbances sourced
communica- services
tion/decision
making
Individual- Less human in- Easy access to Safety in as- More control
ized Traced teraction in da- data, pects crucial on costs and
Data ta completion Preferable data for competi- flexibility in
and integration ordering for- tiveness. this field
End-to-End digital integration
mats
Real-Time Autonomous Increased reli- Security of the Cost effective-
Operating breakdown fix- ability, in- system without ness
Systems ing/replacemen creased effi- additional legal
t systems ciency efforts
Simulation Rapid progress More control More control Cost effective-

and Model- in augmented on the process over data ness
ing of reality Smaller gap More security
products New possibili- between reality especially
and pro- ties and models while innova-
cesses tions tested
Simultane- Progress in ar- More control Safety for pro- Cost effective-
ous plan- ea of digital over data ducer and cus- ness
ning of twin creating More integra- tomer
products and manage- tion between
and produc- ment; construction
tion pro- Progress on and technology
cesses mass customi- areas
zation
Value Add- More autono- Accurate ac- Customer pri- Cost effective-
ed Services mous and actions and reac- vate data safety ness, lower
curate custom- tions price of the
er service products with
systems; bonus services.
Embedded
communication
between pro-
ducer and user
Fig. 5. Cloud computing advantages as growth factor in End-to-End digital integration

reduction. Distributed control has obvious advantage thanks to the local accurate data
analysis. Similar results will refer to the self-optimization. Rapid manufacturing will
benefit from accelerated efforts put on digital twin development. Cloud computing as
mentioned before will not be replaced but stay an important ingredient of hybrid cloud
and edge solutions with less security concerns and pressure on CSP.
In the second field (Fig. 4) - horizontal integration in collaborative networks – edge
computing will make collaborative networks and distributed manufacturing easier in
context of data reliability and quick exchange. The information directly from the
process will shorten the procedure of the network configuration for the given
project/order. The mutual trust issues may be also resolved. Similar effect may be
predicted for Supply-Chain. IoT is the most visible in the research tackling edge
computing impact on Industry 4.0. It is predicted that here will be the most opportu-
nities to be utilized and changing life standard of entire societies.
The third theme package (Fig. 5) - End-to-End digital integration will win thanks to
the more embedded intelligence in the virtualized process chain as well as in simulation
and modelling. Advancement in this steps will make simultaneous product planning
and production process possible.
4 Example of Implementation Concept in Video Data

Streaming
The edge computing solutions are being developed in other fields than manufacturing
but their functionalities and architecture prove that they may easily be adopted and
contribute to the Industry 4.0. The case study shows a concept of innovative edge
computing device dedicated for 180° spherical live video data streaming. The system is
designed to make high quality live video transmissions dedicated for viewers using
AR/VR goggles and share them on media platforms. The core element of this solution
will be device for capturing and transmitting video stream (DfCT) which plays a role of
edge device. The unique computer unit (Figs. 6 and 7) is responsible for processing
signal coming from cameras. The algorithms that are going to be implemented in the
device enable to achieve four times smaller delay in comparison to the solutions
available on the market. It was assumed that maximal value of this parameter is 15 fps.
The designed computer unit will not only be able to quickly capture and share the
pictures from cameras but also provide exceptional quality. The resolution is supposed
to reach 4 K (3840 2160) with minimum 30 fps with the expectation of 21.3 pixels
per one degree. The algorithms implemented will be able to make modifications of the
picture geometry, color field (LUT3D color grading) as well as optics and cameras
imperfections correction. Beside that the video picture will be adopted to the appro-
priate protocols required by the media platform. The important function of the DfCT is
data saving. All the captured video data are saved so beside the online improvements
offline postproduction work can be done. Here is also the point where safety issues are
resolved because only chosen data can be transmitted further to the online platform.
34 J. Gąbka
The edge device described above is a part of a complex solution for multi-camera video
capturing and multichannel transmission servicing live stereoscopic 180° data
streaming for Virtual Reality (Fig. 8). The system consists of the special camera rig
adopted to this kind of transmissions, minimum two cameras for 180° video production
(the system may serve for multi-camera transmissions as well), the edge DfCT video
stream and efficient redistribution platform equipped in universal integration module
enabling third part video streaming.
Fig. 6. DfCT view from the side. Copyright © BIVROST Sp. z o.o.
Fig. 7. General PCB view. Copyright © BIVROST Sp. z o.o.

AR Viewer
Internet
RedistribuƟon
Plaƞorm
Dedicated
camera rig • Capturing
video data
stream
• CorrecƟons
Edge device • Quick, Cloud server
reliable
transmission
• Data storage
Fig. 8. Schematic view of the solution developed for live video streaming.
The system being developed has multiple applications. Its implementation in the
production plant as an integral element of complex, industrial, multi-camera system
enables significant reduction of expenditure on network infrastructure, giving also new
technical possibilities. The predominance of the edge computing approach with DfCT
usage is presented in Fig. 9. Multi-camera system in the case was simplified to eight
cameras but it can be easily expanded to n-points with up to six cameras per each
DfCT.
Camera group I Camera group II
Camera 1 Camera 2 Camera 3 Camera 4 Camera 1 Camera 2 Camera 3 Camera 4

5-10Gbit raw image
5-10Gbit raw image
5-10Gbit raw image
5-10Gbit raw image
5-10Gbit raw image
5-10Gbit raw image
5-10Gbit raw image
5-10Gbit raw image

data
data
data
data
data
data
data
data
Input interfaces Input interfaces

DfCT 1
DfCT 2
Image compression Image compression
Local data storage Local data storage
SelecƟve Access SelecƟve Access
100 Mbps compressed 100 Mbps compressed

videon data videon data
Network hub
Less than 1Gbps mulƟple compressed streams
Centralized server infrastructure
Fig. 9. Schematic view of the multi-camera system equipped in the designed DfCT.
36 J. Gąbka
The input interface of the Device for Capturing and Transmitting data is capable of
receiving 5–10 Gbit of raw image data from the camera controller while the standard
solutions equipped in a network hub at this place can only transmit maximum 1 Gbit of
the raw image data. The data capturing, initial processing, analysis, compression and
storage can be efficiently done locally by DfCT. The problem of costs resulting from
consuming large bandwidth for sending significant numbers of data Gigabits which
was identified while using GigE standard is solved this way. What is more the recorded
material may be filtered locally and only the important interesting image or stream of
image can be send to the centralized server. Simultaneously there is an option to reach
pictures from other cameras or another time period thanks to the picture recording and
compression functions embedded in DfCT. It can be estimated that after initial pro-
cessing only approximately 1 Gbps of data needs to be transmitted to the main server.
In the old standard systems, in comparison, it would be around 8–80 Gbps (depending
on camera connection type), to maintain image quality and performance, in case of
eight cameras operating. The described characteristics all together increase robustness
of the centralized system. The plugins of Device for Capturing and Transmitting data
can execute urgent tasks while waiting for the main system availability. It creates
backup in situations when central system temporarily fails.
The described device delivers also value in other fields. It will enable to get a new
sensations out of watching transmitted events like competitions, conferences, inter-
views. Beside the benefits resulting from the reality effects for the viewer it has great
potential in field of education. E.g. there might be live transmission of a surgery for
different groups of students showing another aspects of the procedure best from their
professional specialty perspective.
5 Conclusions
It is visible that cloud computing widely spread in different application areas met
certain constraints regard time of response, network bandwidth and data safety and
operational service reliability in general. This problems may be resolved with a new
edge computing approach which additionally brings a great opportunities in different
fields connected with Industry 4.0. It is especially considered as an innovation engine
for IoT but can also benefit most of research and consequently application areas dif-
ferentiated within Industry 4.0. The case study of the Device for Capturing and
Transmitting data designed by BIVROST Sp. z o.o. for stereoscopic 180° live video
streaming shows clearly a broad spectrum of implementations that could be considered
in context of manufacturing and related fields like simulations. The edge computing
operating together with cloud computing not only reduces risks but enable to harvest a
new opportunities unattainable before.
Acknowledgments. The project is granted by European Union and financed from European
Regional Development Fund under the Smart Growth Program. The project is implemented as a
part of ‘Szybka Ścieżka’ competition organized by The National Centre of Research and
Development. POIR.01.01.01-00-1111/17.
References
1. Brettel, M., Friederichsen, N., Keller, M., Rosenberg, M.: How virtualization decentraliza-
tion and network building change the manufacturing landscape: an industry 4.0 perspective.
Int. J. Inf. Commun. Eng. 8(1), 37–44 (2014)
2. Liao, Y., Deschamps, F., Loures, E.D.F.R., Ramos, L.F.P.: Past, present and future of
industry 4.0 - a systematic literature review and research agenda proposal. Int. J. Prod. Res.
55, 3609–3629 (2017)
3. Ai, Y., Peng, M., Zhang, K.: Edge computing technologies for Internet of Things: a primer.
Digit. Commun. Netw. 4, 77–86 (2017)
4. Bizanis, N., Kuipers, F.A.: SDN and virtualization solutions for the Internet of Things: a
survey. IEEE Access 4, 5591–5606 (2016)
5. Razzaque, M.A., Milojevic-Jevric, M., Palade, A., Clarke, S.: Middleware for Internet of
Things: a survey. IEEE Internet Things J. 3, 70–95 (2016)
6. Chiang, M., Zhang, T.: Fog and IoT: an overview of research opportunities. IEEE Internet
Things J. 3, 854–864 (2016)
7. Gayathri, M.R., Srinivas, K.: A survey on mobile cloud computing architecture, applications
and challenges. Int. J. Sci. Re. Eng. Technol. 3(6) 2014
8. Mao, Y., You, C., Zhang, J., Huang, K., Letaief, K.B.: A survey on mobile edge computing:
the communication perspective. IEEE Commun. Surv. Tutorials 19, 2322–2358 (2017)
9. Shahzadi, S., Iqbal, M., Dagiuklas, T., Qayyum, Z.U.: Multi-access edge computing: open
issues, challenges and future perspectives. J. Cloud Comput.: Adv. Syst. Appl. 6 (2017)
10. Mell, P., Grance, T.: The NIST definition of Cloud Computing, Version 15, 10 July 2009.
https://www.nist.gov/sites/default/files/documents/itl/cloud/cloud-def-v15.pdf. Accessed 9
Nov 2018
11. Varela, M., Putnik, G., Manupati, V., Rajyalakshmi, G., Trojanowska, J., Machado, J.:
Collaborative manufacturing based on cloud, and on other I 4.0 oriented principles and
technologies: a systematic literature review and reflections. Manag. Prod. Eng. Rev. 9(3),
90–99 (2018)
12. Khan, I., Trzcieliński, S.: Information technology adaptation in Indian small and medium
sized enterprises: opportunities and challenges ahead. Manag. Prod. Eng. Rev. 9(3), 41–48
(2018)
13. Birje, M.N., Challagidad, P.S., Goudar, R.H., Tapale, M.T.: Cloud computing review:
concepts, technology, challenges and security. Int. J. Cloud Comput. 6(1), 32–57 (2017)
14. Gibson, J., Rondeau, R., Eveleigh, D., Tan, Q.: Benefits and challenges of three cloud
computing service models, pp. 198–205. IEEE 2012
15. Jadeja, Y., Modi, K.: Cloud computing - concepts, architecture and challenges. In: 2012
ICCEET, pp. 877–880. IEEE (2012)
16. Intharawijitr, K., Iida, K., Koga, H.: Analysis of fog model considering computing and
communication latency in 5G cellular networks. In: IEEE International Conference on
Pervasive Computing and Communication Workshop, pp. 1–4 (2016)
17. Oueis, J., Strinati, E.C., Barbarossa, S.: The fog balancing: load distribution for small cell
cloud computing. In: 2015 IEEE 81st Vehicular Technology Conference VTC Spring, pp. 1–
6 (2015)
Model Proposal to Evaluate the Quality
of a Production Planning and Control Software
in an Industrial Context
Ricardo M. P. Gonçalves1(&), Maria L. R. Varela1(&),

Ana M. Madureira3, Goran D. Putnik1, and Jose Machado2
1
Department of Production and Systems, University of Minho,
Braga, Guimarães, Portugal
pg32968@alunos.uminho.pt, leonilde@dps.uminho.pt
2
Department of Mechanical Engineering,
University of Minho, Guimarães, Portugal
3
School of Engineering, Polytechnic of Porto (ISEP/IPP), Porto, Portugal
Abstract. The domain of Production Planning and Control, or in a broader

sence Production Management has been deserving a special and increasing
attention by the companies, which intend to continuously achieve better results
through continuous improvement, which also fits in the context of Industry 4.0.
Companies tend to implement management systems with the purpose of
achieving greater competitiveness and, consequently, greater sustainability in
their sector. The selection of the appropriate production management system is a
serious problem for the companies. The main objective of this study is to
support companies in the correct choice of a Decision Support System. The
method used to achieve the proposed objective consists on formulating a model
for comparing functionalities and specifications, where selection of criteria were
also defined and analyzed. Based on a large Company scenario, the model is
applied to three production execution systems: SAP PP (Systems Applications
and Products - Production Planning), Prodsmart and GenSYS.
Keywords: Production management Production management systems

Production planning and control Decision support systems
Comparative analysis
1 Introduction
Society is currently very strict and complex, which makes it difficult for companies to
meet customer expectations, thus increasing the competitiveness and complexity of
markets by giving companies greater flexibility and speed of response, as well as a
rapid take-up [1].
The increase in the number of pieces produced does not necessarily translate into
increased profit, if this increase does not result from the use of more efficient pro-
duction processes that are able to handle the needs of the most diverse products, in
different amounts and moments of time. Therefore, in order to achieve these objectives,
companies must design their production systems so that they can plan, program and

https://doi.org/10.1007/978-3-030-18715-6_4
Model Proposal to Evaluate the Quality of a Production Planning 39
control the entire production process. Thus, it becomes possible to have a detailed
knowledge of the productive processes and their variables, allowing the detection of
opportunities to improve the productive activity and increase the companys’
performance.
Currently, the different hierarchical levels of an industry are well supported by the
tools of Information Technology (IT), which allows the effective supervision and
control of specific subsystems [2–4]. However, an obstacle to the adoption of an
integrated information system stems from the lack of standards and the existence of
various market solutions offered by various suppliers. These solutions, using IT, allow
access to the right information, in the right place, at the right time, and in the right
format [4].
The Production Planning and Control (PCP) is fundamental for improving the
performance of a production system, however, it is evident that even a good PCP
system may not be 100% efficient and not overcome the deficiencies of the design and
organization of the same system, but in many cases dictates its survival or not in the
market. It is common for the PCP not to be able to follow the evolution of production
systems whose complexity is increasing, thus reducing its performance to below sus-
tainable levels, and may often jeopardize the company’s viability [5].
2 Production Planning and Control Systems
A Production Planning and Control System (PPCS) provides information to efficiently

manage material flows, efficiently use people and equipment, coordinate internal
activities with suppliers, and communicate with customers about market needs, along
with interactions among other stakeholders and business partners. The key in this
definition is management need to use information to make smart decisions. A PPCS
does not make decisions or manage operations by itself - managers carry out these
activities. The system provides the support for them to do it sensibly [5].
In production planning and control decision making processes Decision Support
Systems (DSS) hold special importance in situations where the amount of information
available is too much for the intuition of a decision maker who has no support or
knowledge at all through other kind of means or persons and in situations where
process accuracy and optimization are essential [2, 5]. Thus, DSS integrate various
sources of information, and may prove intelligent access to relevant knowledge, while
assisting in the process of structuring decisions, and thus, enable to properly support
the decision making process of managers [6].
A production manager is overwhelmed with data from various sources during a
decision-making process. In this way, it becomes usually a very complicated taks to
extract the relevant information from the vast amount of data to make quick and
assertive decisions. Therefore, the monitoring of the state of the production system, in
conjunction with a well suited DSS, may provide an efficient production management
process [7].
40 R. M. P. Gonçalves et al.
3 System Requirements Specification Model
In a system requirements specification model (SRSM) requirements are goals or con-

straints set by clients and system users that define the various properties of the system.
The software requirements are obviously those among the system requirements that
pertain to software properties. A set of requirements can be defined as a necessary
condition or capacity that the software must possess so that the user can solve a
problem or achieve a goal [8, 9].
In accordance to the general requisites of a SRSM, the proposed model in this work
is conceived to contemplate the following issues: configuration of product structures
(materials); operative ranges; costing; planning; Manufacturing Execution Systems
(MES) [9].
After defining this issues, for being addressed and included in the proposed model,
these were explored to analyze the required and fundamental functionalities for each
issue, as shown in Table 3 set forth in Sect. 4.
In order for the model to be comprehensive for the sector in which it focuses,
several exchanges of impressions and knowledge were made among several consultants
of management software companies, in order to minimize the limitations of the
developed model.
3.1 Selection Criteria and Parameters

Production Management (PM) software is an asset to any company, however, choosing
it is a crucial and compromising decision for your future.
Therefore, due to the complexity of such decision, it is necessary to have an early
study of the functionalities inherent to the company, as well as the definition of
selection criteria and parameters. Such criteria may vary from company to company,
however, most will be similar among them, but with different levels of importance.
There are many different kind of approaches and methods that can be used for sup-
porting a decision making process for selected a best suited software, such as is the case
in this work, about the selection of a PPCS, for instance based on analytic hierarchy
process (AHP) [10] The designed selection criteria and parameters used in our pro-
posed model are shown in Table 1.
These criteria and parameters were defined considering the current scenario of the
industrial sector, namely “Industry 4.0”. This concept gains an increasing weight for
companies since it is an evolutionary trend, where production processes become
increasingly efficient, autonomous and customizable. In this sense, the quality of the
system to be adopted is also an important factor, highlighting, among the main benefits:
cost reduction; greater productivity in development; deliveries more aligned with
business strategies; agility of the production system to respond dynamically and
flexibly to customer requests.
The agility of a production system, according to the creators of this concept, is a
system with resources (technological, human and information) to respond to the
changing needs of the market (flexibility, customers, competitors, suppliers, infras-
tructure, capacity of response, among others) [11–15].
Table 1. Selection criteria and parameters defined.

Criteria Parameters
Cost Determination of cost affection to the adoption of the
system (CS)
Friendly features/interface Appreciation of the software user interface (UI)
Versatility in the planning Contemplation of the number of functionalities of the
module planning module (FMP)
Ease of use Inquiry about their intuitive ability (CP)
Methods of problems solution Appreciation of the ways in which problem solutions were
exposed (PE)
Ability to adopt solutions Contemplation of the ability to generate solutions
automatically (GSA)
Solution change ability Observation of the possibility of adapting or manually
altering the solution generated by the system (ASS)
Local use Use of servers installed in the client company (UL)
Cloud usage Use of servers installed in the company providing the
system (UC)
Integration allow for import and Import and export of data in different formats (IIED)
exportation of data
Permission to integrate with Integration of the system with other software (IOS)
other software
Integration between systems is another relevant factor, and most of these integra-
tions will be centralized in ERP. In this way, it is important that the software guarantees
all communication efficiently and is essential to keep all the technologies aligned in a
robust and reliable platform, capable of achieving the expected results [15].
Regarding the level of importance of each criterion, this varies according to the size
of the company, that is, a certain criterion for a microenterprise may not have the same
importance that it has for a large company. For example, for a microenterprise, the
“Cost” factor will have a much greater weight when deciding which system to adopt
when compared to a large enterprise.
In a percentage scale of 0 to 100, weights (percentage values) were assigned to each
criterion, considering the type of company (micro, small and medium enterprises and
large companies), totaling 100% in each type. The degree of importance for each
selection criterion was also defined in the various company contexts, according to
Table 2, represented in the last column of the table. Such value represents the total
weight of the criterion, in percentage, resulting from the application of Eq. (1).
X ¼ ðPM þ PPM þ PGÞ=3: ð1Þ
Where: “X” represents total weight selection criterion, in percentage; “PM” re-
presents the weight of the selection criterion for microenterprises; “PPM” represented
the weight of the selection criterion for small and medium enterprises; “PG” represents
the weight of the selection criterion for large companies.
Table 2. Weight/importance of defined selection criteria.

Selection criteria and parameters Microcompany Small & medium Big Total
business companies weight
Cost (CS) 20 15 10 15
Friendly features/interface (UI) 10 5 5 7
Versatility in the planning module 5 5 5 5
(FMP)
Ease of use (CP) 15 10 5 10
Scheduling methods (EP) 10 10 10 10
Other problem solving methods 10 15 15 13
(PE)
Solution change capacity (ASS) 5 10 15 10
Local use (UL) 10 5 5 7
Cloud usage (UC) 5 5 10 7
Integration allow for data import 5 10 10 8
and export (IIED)
Permission to integrate with other 5 10 10 8
software (IOS)
It is notorius that for a large company, the most important criteria are the “Capacity
to adopt solutions” and “Capacity to change solutions”.
It is also possible to verify that the factor “Cost”, due to its sensitive nature in the
phase of adopting software, is the criterion with greater total weight in relation to the
other criteria. Next, we highlight the “Capacity to adopt solutions”, that is, the ability to
generate solutions automatically, an increasingly important factor in “Industry 4.0”,
which is basically ruled through automation.
3.2 Correlation Between Selection Criteria

Regarding the selection criteria defined above, the possible existence of validations is
evident, since some criteria may be influenced or interdependent of others. By
observing the “Cost” factor, this can corelate with “Versatility in the planning module”,
and the cost may change due to the number of functionalities in the planning module. In
turn, “Versatility in the planning module” can be correlated with “Ease of use”, and the
more functionalities the system adopts, the greater the complexity in terms of its use.
The “Cost” factor can also be correlated with “Solution Capability” and “Solution
Change Capability”, and enabling these capabilities is a fundamental aspect of the
current competitive requirements of “Industry 4.0”. Thus, the factor “Cost” may
fluctuate according to the existence or not of these capacities.
Table 3. Model of features and specifications and results obtained.

Features & specifications SAP PP GenSYS Prodsmart
Configuration of product structures (Materials):
Possibility of product structures contemplating x x x
alternatives
The need for articles to contemplate the possibility of x x x
substitution
Operative ranges
Possibility of the operative ranges contemplating x x x
alternatives
Possibility of aggregation of products to facilitate x x
planning
Costing
Disaggregation by type of materials x x x
Disaggregation by type of waste x x
Subcontracting of products x
Collaborator/operation runtime x x x
Runtime machine x x x
Distribution of departmental costs (general x
manufacturing expenses) by specific criteria
Cost per product x x x
Cost per product/line x x
Cost line x x
Productivity collaborator x x x
Productivity line x x
Productivity line/product x x
Planning
Relation operation/execution time x x x
Possibility of automatic production optimization x x
Possibility of optimization simulation x x
Graphical display of loads per line x x x
Graphical display of loads per operation x x x
Graphical display of loads by operator x x x
Graphical display of overloads x x x
View in calendar monitored operations by x x x
line/operation/operator
MES
Integrated add-on with ERP remainder x x
Graphical and intuitive graphical environment x x x
Web based technology or APP x x x
Optical scanning (barcode) x x x
Connecting to barcode printers x x x
Parametric and multi-language product design labels x x x
(continued)
Table 3. (continued)
Features & specifications SAP PP GenSYS Prodsmart
Possibility to connect with other equipment x x
Possibility of selecting manufacturing orders to be x x x
executed on each production line
Association of the employee to the line and the x x x
operation and exchanges of lines and operations
Typical stops reports x x x
Quality checks by quantity, by time, by operation and x x x
combinations
Association of files to the operations (videos, PDF, x x x
images)
Association of files with manufacturing orders x x
Association of files to articles to be produced x x x
Possibility of selecting and viewing videos and x x x
previous files
Quantities produced x x x
Time spent per line x x x
Time spent per operation x x
Materials consumed x x x
Real-time line state x x
Real-time order status x x x
Real-time general productivity x x
Real-time line productivity x x
Real-time collaborator productivity x x
Delays compared to original real-time planning x x x
Using a kanban system x x
Ability to handle high diversity and quantity of articles x x x
Real-time analysis of production status x x x
Connection with machines x x
OEE (Overall Equipment Effectiveness) x x
Analysis and representation of non-productive x x x
operations
4 Results and Discussion
The obtained information regarding the proposed model, based on main defined
functions and specifications, is shown in Table 3, as well as the obtained data about the
criteria and selection parameters used, presented in Table 4. These data was obtained
through surveys and semi-structured interviews conducted directly with company
employees of the respective software, which were employees who hold senior posi-
tionsat the corresponding companies or are expert users.
Table 4. Practical systems analysis

Selection SAP PP Prodsmart GenSYS
criteria/softwares
Cost € 300,000 + € 1500 per € 1,599 month by 10 € 200,000 +
user + € 100 per hour of users + € 99 month per maintenance
assistance extra user contract
Friendly Yes Yes Yes
features/interface
Versatility in the Yes Not specified Yes
planning module
Ease of use Not very intuitive Yes Not very
intuitive
Troubleshooting Yes Not specified Yes
methods
Ability to adopt Yes Not specified Yes
solutions
Solution change Yes Not specified Yes
ability
Local use Yes No Yes
Cloud usage Yes Yes No
Integration allow Yes Yes Yes
for data import
and export
Permission to Yes Yes Yes
integrate with
other software
By analyzing the data provided, there are visible discrepancies in terms of func-
tionalities among the systems, however, SAP software stands out for its comprehen-
siveness, covering all the requirements of the developed model.
The GenSYS and Prodsmart software do not include some functionalities, and in
general, do not include “Subcontracting of products” and “Distribution of departmental
costs (overhead costs) by specific criteria”.
There are some criteria where it was not possible to obtain information, however, as
the Prodsmart system does not calculate the material requirement planning (MRP) and,
since the evaluation parameters are linked to the MRP, it is assumed that it does not
comply with the criteria without information.
Considering the factor “Cost”, a factor of greater general importance, this presents a
high complexity of calculation since the companies’ present different and differentiated
needs. However, some values regarding software implementation costs were calculated
and, from these intervals, the cost for a large company was estimated.
Regarding the duration of the implementation and testing phase, for both SAP PP
and GenSYS software, this can exceed 12 months, however, Prodsmart software only
requires 2 months for its implementation. Considering all the data, Prodsmart is
expected to have a significantly lower cost than the other systems.
Given the weight of the selection criteria for a large company, these are of major
importance and are satisfied with SAP PP and GenSYS software. On the other hand,
observing the lower values, SAP PP software and GenSYS proved to be equivalent,
however, the Prodsmart system stands out in terms of “Ease of use”.
In general, the Prodsmart system, despite being a management system, for analysis
and optimization of industrial production processes, it does not meet the needs of a
large company, and is a more appropriate system for small to medium enterprises. The
remaining software is robust and capable of increasing the competitiveness of a large
company, providing optimized and controlled production as well as other inherent
positive aspects for management processes support. Thus, the “Cost” factor will be
preponderant for decision-making regarding the system to use, and in this regard,
GenSYS software stands out being the best suited one.
5 Conclusion
The use of robust and complete software in prouction management that correctly
terminates costs and enables a clear view of business productivity is currently con-
sidered to constitute a fundamental pillar of process automation in the industrial
companies.
The system developed by GenSYS, after analyzing the results obtained, is the most
advantageous solution for the production sector with the underlying companies and
software analysed, being capable of satisfying the needs of a large company. Moreover,
is stand out to be an intelligent, flexible, powerful and fully integrated solution capable
of creating solutions to guide a manager thorugh decisions based on a production
planning and control system. However, for all its features, utilities and efficiency,
GenSYS is currently unable to compete closely with the SAP system. SAP is so
possant and is so rooted in the market that it becomes difficult for other software to
compete with it, and there is even the general idea a existing corelation between the
quality of management in a company through the use of this system that turns out to put
SAP as a softwar in the market that acts as a kind of “brand image” of quality,
efficiency and sovereignty, which is very hard to overcome by other production
management software.
Acknowledgments. This work has been supported by FCT Fundação para a Ciência e Tec-
nologia, under the Project with reference PEst2015-2020.
References
1. Shuai, J.J., Su, Y.-F., Yang, C.: The impact of ERP implementation on corporate supply
chain performance. In: 2007 IEEE International Conference on Industrial Engineering and
Engineering Management, pp. 1644–1648. IEEE (2007)
2. Varela, M.L.R., Aparício, J.N., Silva, S.C.: A Web-based application for manufacturing
scheduling. In: IASTED International Conference on Intelligent Systems and Control,
pp. 400–405 (2003)
3. Campanella, G., Ribeiro, R.A., Varela, L.R.: A model for B2B supplier selection. Adv.
Intell. Soft Comput. 107(6), 221 (2011)
4. Glavan, M., Gradišar, D., Strmcnik, S., Mušic, G.: Production modelling for holistic
production control. Simul. Model. Pract. Theory 30, 1–20 (2013)
5. Varela, M.L.R., Ribeiro, R.A.: Distributed manufacturing scheduling based on a dynamic
multi-criteria decision model. In: Recent Developments and New Directions in Soft
Computing, Springer, pp. 81–93 (2014)
6. Druzdzel, M.J., Flynn, R.R.: Decision Support Systems. Und Inf. Retr. Syst., 461–72 (2011)
7. Jovan, V., Zorzut, S., Gradišar, D., Dorneanu, B.: Closed-loop production control of
polymerization plant using production KPIs. IFAC Proc. 40(18), 553–558 (2017)
8. Arrais-Castro, A., Varela, M.L.R., Putnik, G.D., Ribeiro, R., Dargam, F.C.: Collaborative
negotiation platform using a dynamic multi-criteria decision model. Int. J. Decis. Support.
Syst. Technol. 7(1), 1–14 (2015)
9. Zhong, R.Y., Dai, Q.Y., Qu, T., Hu, G.J., Huang, G.Q.: RFID-enabled real-time
manufacturing execution system for mass-customization production. Robot. Comput.-Integr.
Manuf. 29(2), 283–292 (2013)
10. Vaidya, O.S., Kumar, S.: Analytic hierarchy process: an overview of applications. Eur.
J. Oper. Res. 169(1), 1–29 (2006)
11. Sherehiy, B., Karwowski, W., Layer, J.K.: A review of enterprise agility: concepts,
frameworks, and attributes. Int. J. Ind. Ergon. 37(5), 445–460 (2007)
12. Trojanowska, J., Dostatni, E.: Application of the theory of constraints for project
management. Manag. Prod. Eng. Rev. 8(3), 87–95 (2017)
13. Rewers, P., Trojanowska, J., Diakun, J., Rocha, A., Reis, L.P.: A study of priority rules for a
levelled production plan; In: Hamrol, A., Ciszak, O., Legutko, S., Jurczyk, M. (eds.)
Advances in Manufacturing. Lecture Notes in Mechanical Engineering, pp. 111–120 (2018)
14. Trojanowska, J., Kolinski, A., Galusik, D., Varela, M.L.R., Machado, J.: A methodology of
improvement of manufacturing productivity through increasing operational efficiency of the
production process. In: Hamrol, A., Ciszak, O., Legutko, S., Jurczyk, M. (eds.) Advances in
Manufacturing. Lecture Notes in Mechanical Engineering, pp. 23–32 (2018)
15. Varela, M.L.R., Putnik, G.D., Manupati, V.K., Rajyalakshmi, G., Trojanowska, J.,
Machado, J.: Collaborative manufacturing based on cloud, and on other i4.0 oriented
principles and technologies: a systematic literature review and reflections. Manag. Prod.
Eng. Rev. 9(3), 90–99 (2018)
Increasing the Production Productivity
with Artificial Bee Colony Optimisation
Method
Danijela Pezer(&)
University Department of Professional Studies, University of Split,

21000 Split, Croatia
dpezer@oss.unist.hr
Abstract. In this paper, the tool path length minimisation or reduction of tool
path total time was considered. The main goal is optimisation of tool path length
on the selected technological task, where is need to drill a large number of holes
and Artificial Bee Colony (ABC) optimisation method was used. The results,
(achieved by criteria of minimum tool path) leads to saving of the technological
time and reducing the total costs of production. Proposed algorithm gives the
sustainable results, and it is reliable for the use. The solution achieved by the
ABC algorithm was implemented in the MATLAB program and for validation
of its performance it was compared with Ant Colony Optimisation (ACO) al-
gorithm, CAMConcept software and with the result achieved by manual pro-
gramming. The drilling simulation was performed using the EMCO WinNC
educational program for the Sinumerik 840D Mill control unit.
Keywords: Optimisation Artificial Bee Colony Ant Colony Drilling
1 Introduction
The Artificial Bee Colony is an optimisation algorithm which belongs to the group of
bee algorithms and it was introduced [1] by Karaboga 2005. The algorithm, based on
the optimisation of a bee colony, is a stochastic P-metaheuristics that belongs to the
class of swarm intelligence algorithms. Many studies based on different behaviors of
bee colonies have been developed to solve complex combinatorial or continuous
optimisation problems [2].
The ABC algorithm is easier to implement than the other swarm intelligence
algorithms (such as Ant Colony Optimisation algorithm), and it has been applied to
solve many problems, such as a flow shop scheduling problem [3], clustering problem
[4] and Job Shop Scheduling Problem [5]. In previous years, the ABC algorithm has
been modified to solve combinatorial discrete optimisation problems such as a Trav-
eling Salesman Problem [6–9].
This paper proposes the use of ABC and ACO algorithm (because both methods
belongs to the same class of swarm intelligence algorithms) for the analysis of the
problem based on problem similar to the Traveling Salesman Problem (TSP), or
troubleshooting of tool path optimisation at drilling sequence. The basic assumption at

https://doi.org/10.1007/978-3-030-18715-6_5
Increasing the Production Productivity with Artificial Bee Colony 49
the TSP problem is that the salesman should return to the starting point – city (drilling
hole in this paper) from which tour started, which is called a closed tour, otherwise we
are dealing with an open tour, which will be considered in this paper.
Since the efficiency of production is of importance for each manufacturing process,
the tool path optimisation in the drilling process is essential, because it leads to increase
of productivity and saving of the production costs. This particularly applies to the
process of drilling a large number of holes (for example on a pipe wall in a heat
exchanger), where the tool that performs the drilling operations must visit a significant
number of places in order to perform drilling. In that case, the production is dependent
on the time required for drilling. Ultimately, the goal in mass production is to produce
the high-quality products at reasonable prices with the reduced production costs and
increased production efficiency.
2 Methods
2.1 Artificial Bee Colony (ABC)
The ants leave pheromone trails, the birds are aligned with the neighbors when flying,
and the bees - dance. The bees communicate with the “waggle dance” whose meaning
was deciphered by Karl von Frisch. The waggle phase is crucial to transfer information
to the other bees. The bees with waggle dance indicates three things: source direction,
source distance, and source quality. The role of bee colony is to find a rich food source
to achieve the maximum nectar amount. Behavior of food search (the nest selecting,
food foraging) and behavior in the bee colony community are the main activities in bee
colony life that attract scientists to create optimisation algorithms. Artificial Bee Col-
ony is divided into three groups:
• Employed bees - explores the environment of a source that has previously visited
(bees associated with certain food sources).
• Onlookers - in the transmitted sense observe bees on the dance floor while per-
forming the bee’s dance and make the decision which bee will they follow.
• Scouts - conduct a random search space (randomly looking for a food source).
2.2 Artificial Bee Colony Algorithm for Drilling Sequence Planning

The problem to be considered in this paper, the determination of the optimum sequence
of drilling, based on the traveling salesman problem (symmetrical TSP), where each
hole is drilled only once, and at the same time the coordinates of the defined holes are
known. The goal is to find the minimum tool path length dij, which represents the
Euclidean distance and for the plane problems, it is expressed by the equation [10]:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2ffi
dij ¼ xi xj þ yi yj ð1Þ
50 D. Pezer
The limitations of the model of problem:

• selection of the initial holes depends on the random algorithm selection
• after the drilling of last hole, the tool remains in its current position
• the tool wear was not considered
• drilling a hole of the same diameter (and depth).
The fitness function, or the total minimum tool path is defined as:
X
min dij xij ; ð2Þ
ði;jÞ2E
with degree constraints:

X X
xij þ xji ¼ 2; i2V ð3Þ
j2V:ði;jÞ2E j2V:ðj;iÞ2E
With the TSP problem, it is necessary to ensure a continuous tour of holes, i.e. that
the edges/arcs are associated with the initial hole all the way of the tour, and each hole
has to be visited. In order to ensure that, the limitation is set to eliminate all possible
subtours (the solution containing degenerative visits/tours between the central vertex
that are not associated with the initial hole).
Subtour elimination constraints:
X
xij jSj 1; S V; 3 jSj n 3 ð4Þ
ði;jÞ2E:fi;jg2S
where S is a subset of vertices of G.

Integrality constraints:
xij ¼ 0 ili 1; ði; jÞ 2 E ð5Þ
The expression (3) ensures that each hole is drilled only once, and according to the
second limitation, the expression (4), the subtours are not allowed, and the possibility
that the solution has more than one path was eliminated.
In order to solve the TSP problem using the ABC method, the food source rep-
resents a tour, while nectar (fitness) of food sources corresponds to the tour quality. The
set of food sources was randomly generated in employed bee phase. The nectar of the
collected food sources is determined by the fitness function.
2.3 Solution Construction for ABC Method

Employed bees: Bees in nature that are related to a source also observe the immediate
environment of that source in the hope of finding a better source than the current one.
This process is mapped to the computer, not by comparing the entire environment, but
by randomly generating a solution compared to the current one. If the new generated
solution is better than the old one, then the artificial bee remembers a new source,
otherwise it is tied to the old source. Below is given the basic equation of the bee
algorithm, which except the employed bees also use and onlooker bees. By this
equation, based on the old solution (vector xi ) generates a new solution (vector vi ) for
all bees [11]:
vij ¼ xij þ Uij ðxij xkj Þ; ð6Þ
where k 2 f1; 2; . . .; N g and j 2 f1; 2; . . .; Dg are randomly selected index and D is the
dimensionality of the problem. The greedy method remembers only a better vector
between xi and vi .
xi is the solution associated to the i-th employed bee; xij is j-th element of solution xi .
vi is the new solution produced based on xi ; vij is j-th element of solution vi ;
Uij is random real number between [−1, 1].
Onlookers: Further, the bees that are not employed look at the results of employed
bees. It is logical that the bees that have an information of more quality solutions attract
more “followers”. This principle is realized by the method of simple selection “roulette
wheel selection” [11]:
fiti
Pi ¼ PN ; ð7Þ
i¼1 fiti
where Pi indicates the probability that the bee will accept the solution i as her own.
fiti – fitness function value of the solution i; N – number of food sources.
Onlookers after the decision which solution will they accept, conduct the same
procedure as employed bees (expression 6).
Scouts: Scouts (except in initialisation) are the minority in the colony, compared to
the other two groups. The need for scouts occurs only when some solution is rejected,
so new, independent of the previous solutions is needed. Then, a new scout is intro-
ducing, which finds a new solution within the given boundaries – vector d for the
lower, and vector g for the upper bound [11]:
xij ¼ dj þ rand½0; 1 ðgj dj Þ; ð8Þ
The same expression was used for generating the initial population of N solutions.
Scouts are created, after leaving the solutions, if necessary.
By combining the above-described behavior we get an ABC algorithm that com-
bines four types of selection processes: local selection (expression 6), global selection
(expression 7), randomly selection (expression 8) and greedy selection (expression 6)
which is manifested by the greedy choice between the old and the new solution.
52 D. Pezer
The proposed ABC algorithm is implemented in a loop with repeating three basic
steps, each step being responsible for one type of bee. The procedre of solution con-
struction can be shown in several steps:
Step 1. Define the problem, i.e. create a mathematical model of function to deter-
mine the total tool length path.
Step 2. Initialisation phase: Setting parameter values: number of variables, lower
and upper bound, number of population size (food sources number - SN), number of
bees (employed and onlooker bees), the limit value and set the maximum number of
iteration. In the initialisation phase, the set of food sources was randomly selected
by the bees and the nectar amounts were determined.
Step 3. Employed bee phase: Sending the employed bees to their food sources
(source in their memory) and evaluating the nectar amount (after finding a new food
source, its fitness is calculated, and greedy choice between new food source and the
old one was applied). After that, the employed bees sharing those informations with
onlooker bees by dancing in the hive.
Step 4. Onlooker bee phase: Selection of food sources by onlooker bees (based on
information of nectar, carried by employed bees through the waggle dance) and
evaluation of the nectar amount from food source. With an increase the nectar
amount from food source, the probability of selecting that food source also
increases. For this purpose, a technique selection based on fitness (roulette wheel
selection method) was used.
Step 5. Scout bee phase: Randomly determining a new food sources by a scout bees
as a replacement for the abandoned food sources, or if the nectar amount of the new
food source is larger than previous one in their memory, they remember a new
position and forget the previous one. The selection is controlled by a control
parameter called “limit”. In the scout bee phase, the bees whose solution can not be
improved through a predetermined number of attempts, known as the “limit”,
become scouts and their solutions are abandoned.
Step 6. The process is repeating until one of the stop conditions is met, in this case
when the maximum number of iterations is reached.
Step 7. By stopping the process (in step 6), the best achieved solution is shown, i.e.
a sequence display of the possible drilling order according to the assigned number
of holes.
The following pseudo-code can be achieved by the above described method.
Pseudo-code for ABC algorithm

Initialisation:
FOR = 1 TO
Initialise the vector according to the expression (8)
END FOR
cycle = 1
REPEAT
Employed bee phase:
FOR EACH employed bee ( )
Calculate based on according to the expression (6)
IF quality ( ) > quality ( ) THEN =
END FOR EACH
Onlooker bee phase:
FOR = 1 TO
Calculate the probability from according to the expression (7)
END FOR
FOR EACH onlooker bee ( )
Select = (with probability )
Calculate based on according to the expression (6)
IF quality ( ) > quality ( ) THEN =
END FOR EACH
Scout bee phase:
FOR EACH abandon solution ( )
Generate a new random solution according to the expression (8)
END FOR EACH
IF MaxSolution (cycle) > MaxSolution THEN
MaxSolution=MaxSolution(cycle)
cycle=cycle+1
IF (cycle= ) THEN STOP
END
At the population-based ABC algorithm, a food source represents the possible

solution of the optimisation problem (tool path sequence of hole drilling), while nectar
amount of food source corresponds to fitness (tool path length) of the associated
solution. In the proposed algorithm, the number of employed bees is equal to the
number of onlooker bees, while the number of food sources is equal to the half number
of employed bees. While onlookers and employed bees conduct the process of
exploitation in the search area, scouts control the research process. Thus, the ABC
system combines local search methods, conducted by employed bees and onlooker
54 D. Pezer
bees surveyed, with global search methods managed by scouts, trying to balance the
research and exploitation process.
2.4 Ant Colony Optimisation (ACO) Method

Ant Colony Optimisation (ACO) is also classified as swarm intelligence, inspired by
the behavior of ants during the search for food in nature.
Ant Colony Algorithm imitate the movement of ants in the nature, during the search
for the shortest route to the food, or finding the optimal solution for a given problem.
Since ants haven’t well-developed eyesight, their mutual communication is based on
pheromone trails. The amount of pheromone on the ground determines the probability
to be followed by other ants.
The basic algorithm structure [12] of Ant Colony is presented by Algorithm 1.
Algorithm 1. General algorithm for ACO
Initialise the pheromone trails;
Repeat
For each ant Do
Solution construction using the
pheromone trail;
Update the pheromone trails:
Evaporation;
Reinforcement;
Until Stopping criteria
Output: Best solution(s) found
The steps of the algorithm work [12]:

Step 1. Initialisation of pheromone trails (on each way is set the amount of pher-
omone that is equal to the reciprocal value of the path length - found by using the
greedy algorithm)
Step 2. Route selection - an ant selects the next hole from the group of holes that
have not yet been drilled. Since that is a fully connected graph, from each hole it
can be reached to any remaining hole with the probability:
saij gbij
Pi ¼ P ; ð9Þ
k2N saik gbik
Pi;j - the probability of path choosing from hole i to hole j

N - a set of holes that haven’t been drilled yet
si;j - the intensity of the pheromone on the path from hole i to hole j
gi;j - the value of heuristic function on the path from hole i to hole j
a - constant (pheromone trail)
b - constant (visibility between holes).
For a = 0 algorithm acts like Greedy algorithm.
For b = 0 the search is based only on the pheromone trail.

Step 3. Updating the pheromone trails (occurs in two phases):
• Evaporation phase
The pheromone trails are updated for all paths using relation:
sij ¼ ð1 qÞsij ; 8i ; j 2 ½1; n ð10Þ
where
q - evaporation constant, q 2 ½0; 1.
Evaporation is possible to apply after each ant constructs a solution or after each
iteration in which the solution constructed by all ants (as is the case in this paper).
• Reinforcement phase
The pheromone trail is updated according to the last constructed solution and can be
applied after each step of the heuristics construction for each ant, after an each ant
constructs a solution or after all ants construct a solution in which there are several
strategies:
– reinforcement based on quality – the value of pheromone is updated according
to the k best solutions (k = 1:m; m – number of ants) with value proportional to
the quality of solutions:
sip ðiÞ ¼ sip ðiÞ þ D; 8i 2 ½1; n ð11Þ
where
D ¼ 1=f ðpÞ.
– reinforcement based on the range – the k best solutions updating an appropriate
components of trails to the amount which depends on their ranking
– reinforcement based on a worst solution – the worst found solution reduces the
appropriate parameters of trails
– elitistic reinforcement – increasing the value of pheromone for the best solution
found until then.
• Step 4. By stopping the process, the best accomplished solution, or sequential
representation of possible order of execution of drilling (according to the assigned
number of holes), was displyed. Termination criterion is defined by the number of
iterations.
2.5 Ant Colony Optimisation Algorithm for Drilling Sequence Planning

Ants in the nature (in a simple manner with joint forces), perform complex tasks such
as transportation of food and finding the shortest path to the food source.
The ACO algorithm mimics this principle using a very simple communication
mechanism by which an ant colony can find the shortest path between two points.
Since ants do not have a well-developed sight, during the food search, they leave on
the ground chemical traces - pheromones. Pheromones are actually fragrant and volatile
substances, and their role is to lead other ants toward the target point. The greater the
amount of pheromones on a particular path, the greater the chance that the ants will choose
that path, the default ant chooses the route based on the amount of pheromone scent.
56 D. Pezer
Pheromones evaporate over time (evaporation process) and the amount that was left
by one ant depends on the amount of food (reinforcement process). When ants are faced
with an obstacle, the probability of selecting the left or right way is equal for each ant.
If the left path is shorter than the right, and requires less travel time, an ant will finish
the travel leaving a greater level of pheromones on that path. The more ants choose the
left path, the higher is the pheromone trail.
Algorithm 2 [12] shows the Ant Colony algorithm for sequential hole drilling based
on the Traveling Salesman Problem.
3 Results
The ABC algorithm is used for the optimisation of drilling sequence, and has been
tested for numerous numbers of holes (7, 28 and 72). As a defined technological
problem, a prismatic workpiece with 72 holes was used. It was necessary to determine
the minimum tool path, by drilling of 72 holes, and the minimisation of tool path length
was achieved by using the MATLAB software. The results of ABC algorithm were
compared with the testing results of the ACO algorithm (because the ACO method also
belongs to the swarm intelligence group), and also with the results achieved by
CAMConcept software and manual programming (by using the WinNC software). The
initial population for both methods (ABC and ACO) was generated randomly, and
starting point of drilling was also selected randomly.
The first testing of algorithms was performed on examples with a smaller number of
holes (No. h 7 and 28), in order to compare the efficiency of ABC and ACO algorithms.
The same population number and the number of iterations are used, although it is
known that these control parameters have a significant effect on the work of the
algorithm.
The following parameters were used for testing: Number of population (Npop) for
value 10 and 50 in combination with Maximum number of iterations (MaxIter) for
values 100 and 1000, while other parameters of the ACO algorithm correspond to the
parameters of previous author’s researches [13]. The algorithm for each combinations
was started 15 times, and obtained results are shown in Table 1.
Table 1. Comparison of the results achieved with ABC and ACO algorithm for the same
parameters
No. h ABC algorithm ACO algorithm
7 Npop MaxIter Best cost, CPU time, s Npop MaxIter Best cost, CPU time, s
mm mm
Iteration Iteration
10 100 131,531 0,7693 10 100 131,531 6,6385
4th 1st
50 1000 131,531 16,4812 50 1000 131,531 61,6068
5th 1st
28 10 100 162,5438 0,4642 10 100 84.7458 7,0467
97th 5th
50 1000 84,7458 17,0460 50 1000 74,7487 83,6343
772th 1st
Increasing the number of iterations directly affects to the algorithm runtime, as well
as the quality of obtained solutions. A larger number of iterations gives a better solution
but it also increases the algorithm runtime. It is also necessary to determine the optimal
number of iterations by which the algorithm will find a satisfactory solution in a
relatively short time (or in a real-time), and which would be satisfying (acceptable). In
addition to defining the number of iterations, the selection of the population size (the
number of bees) is essential, with whose increase we directly affect to the quality of
obtained solutions and reduce the probability of premature convergence to a local
optimum (and also increase the execution time).
58 D. Pezer
The next test was performed for the defined technological problem with 72 holes.
Population size and the number of iterations are determined experimentally.
A good performance of the ABC algorithm was significantly affected by these
parameters: number of population size (food sources number SN), number of bees
(employed and onlooker bees), the limit value and maximum number of iterations.
The algorithm was started 50 times, using different combinations of mentioned
parameters. According to the algorithm testing with the different combinations of the
above mentioned parameters, for a given problem, the following parameters were
determined as optimal: SN = NB/2, NB = 400, the maximum number of iterations
maxIter = 20 000, by which a minimum total tool path length in the amount of
223,9877 mm (in 11 824th iteration) was achieved. The program was performed on a
laptop Acer Aspire 5741G, i5 430 M CPU, 2.2 GHz, 3 GB of RAM.
ACO algorithm was running several times with different combinations of the
population size and parameters such as a, b and q, in order to determine the parameters
of the algorithm which providing the best solution. Parameter a affects to the impor-
tance of pheromone trail, b represent the importance of visibility (hole distance), while
q is the evaporation factor.
According to the algorithm testing, with the different combinations of the above
mentioned parameters, for a given problem the following parameters were determined
as optimal: a = 1, b = 4, q = 0,45, Q = 1, nAnt = 35, the maximum number of iter-
ations maxIter = 1000, with which a minimum total tool path length in the amount of
223,2434 mm (in the 891th iteration) was achieved. Increasing the number of holes
increases the complexity of the problem, and thus extends the running time, for both,
ACO and for the ABC algorithm.
The total tool path length achieved by the CAMConcept software is 778,2044 mm.
The drawing of prismatic workpiece was imported from the CAD program, and the
holes have been selected for drilling to enable generating the tool path.
The total tool path length achieved by the WinNC software (manual programming)
and defined by the user is 230,1048 mm.
Figure 1 shows the optimal tool path (green start point) achieved by the ABC
algorithm (1.a), ACO algorithm (1.b), CAMConcept software (1.c) and manual pro-
gramming (1.d).
Fig. 1. Optimal tool path achieved by various methods
Table 2 gives an overview of the best solutions achieved by ABC and ACO
methods, but also with different parameters. For example, the ACO with the number of
iterations 1000 and the number of population 35 due to multiple algorithm start
achieves the best solution in the amount of 223,2434 mm, but with a significantly
shorter time (*142 s in the 891th iteration) compared to the ABC algorithm (according
to Table 2) which required *845 s to achieve the solution in the amount of
223,9877 mm in 11 824th iteration.
Table 2. The results achieved with ABC and ACO algorithm, CAMConcept software and
manual programming
Method MaxIter Npop Best cost, mm CPU time, s
ABC algorithm 20 000 200 223,9877 1429,3481
ACO algorithm 1000 35 223,2434 159,5632
CAMConcept software — — 778,2044 —
Manual programming — — 230,1048 —
From the shown results we can concluded that the ACO algorithm finds a satis-
factory solution with the smaller number of iterations compared to the ABC algorithm,
but the execution time of one iteration is much longer in relation to the ABC algorithm.
Considering the relatively small number of iterations required for the efficient work of
the ACO algorithm in relation to the proposed ABC algorithm, another testing for the
same number of iterations and population number for the above mentioned methods
(maxIter = 1000, Npop = 35) was performed. The ABC algorithm requires 12,9609 s
to pass through 1000 iterations with the optimum tool path 455,5742 mm (in the 829th
iteration), while ACO algorithm requires 159,5632 s. However the solution achieved
by the ACO algorithm (223,2434 mm) is significantly closer to the optimal, which
confirmed [11] that the choice of parameters significantly affects to the results, even
when it comes to algorithms which are classified into the same group. Figures 2 and 3
provides a comparison of the impact of the iterations number in the relation to the
objective function (the best obtained solution).
The speed of convergence for The speed of convergence for

ABC algorithm ACO algorithm
2000
300
Best cost, mm
1500
Best cost, mm
280
1000
260
500 240
0 220
0 10000 20000 0 500 1000
Iteration Iteration
Fig. 2. The speed convergence of ABC Fig. 3. The speed convergence of ACO
algorithm for drilling sequence problem algorithm for drilling sequence problem
60 D. Pezer
The Figs. 2 and 3 shows that the ACO algorithm converges faster with a quite
smaller number of iterations, but the results are slightly different, compared to the
proposed ABC algorithm.
4 Conclusion
According to the given technological task, the ABC algorithm achieved a suboptimal
solution with a slight difference in relation to the ACO algorithm, although ABC use
less control parameters. For the successful algorithm performance, parameters selection
is very important. Especially, a selection of the number of population and the maxi-
mum number of iterations (for both ABC and ACO), as well as the limit value in the
ABC algorithm are important, while parameters of ACO algorithm (related to the
pheromone trail and visibility and the evaporation factor) also significantly influence.
The ABC algorithm has a good global search capability, and due to its simplicity of
implementation in relation to the other algorithms, it efficiently finds the optimal or
satisfactory solution within a reasonable time. Since the ABC algorithm finds a
potential solution by random selection from the generated population, the possibility of
staying trapped in local optimum increases. For this reason, scout bees are introduced,
which enable the replacement of the existing solution with a new randomly obtained
solution. In that way, the diversity of the population is encouraged and enables the
search of new - potentially the best solutions. We can say that employed and onlookers
bees are conducting local search in the space of the solution, while scouts conduct a
global search.
The solution achieved with ABC and ACO algorithms are more effective than the
solution achieved with CAMConcept software. The solution achieved by manual
programming gives a satisfactory solution, however it is reduced to a free assessment of
the user, and the effectiveness depends on the type and complexity of the problem.
Further Research
Since, the ABC algorithm is good in exploration and weak in exploitation, there is a
space for algorithm modification to improve its performance, especially in order to
avoid local optimum and accelerate the speed convergence. The continuation of the
research aims to modify and implement algorithms from swarm intelligence group, and
its application to similar problems encountered in real world production, and also to the
direction of the parameter optimization (which significantly influences to the final
solution, and in that sense ultimately lead to even more favourable results of the
proposed ABC and ACO algorithms).
References
1. Karaboga, D.: An idea based on honey bee swarm for numerical optimisation (Technical
report-Tr06, October 2005), Erciyes University, Engineering Faculty Computer Engineering
Department Kayseri/Türkiye (2005)
2. Bitam, S., Batouche, M., Talbi, E.G.: A taxonomy of artificial honeybee colony
optimisation. In: International Conference on Metaheuristics and Nature Inspired Comput-
ing, META 2008, Hammamet, Tunisia (2008)
3. Pan, Q.K., Tasgetiren, M.F., Suganthan, P., Chua, T.: A discrete artificial bee colony
algorithm for the lot-streaming flow shop scheduling problem. Inf. Sci. 181(12), 2455–2468
(2011)
4. Zou, W., Zhu, Y., Chen, H., Sui, X.: A clustering approach using cooperative artificial bee
colony algorithm. Discrete Dyn. Nat. Soc. 16 p. (2010)
5. Chong, C.S., Low, M.Y.H., Sivakumar, A.I., Gay, K.L.: A bee colony optimisation
algorithm to job shop scheduling simulation. In: Proceedings of the Winter Conference,
Washington DC, pp. 1954–1961 (2006)
6. Karaboga, D., Gorkemli, B., Ozturk, C., Karaboga, N.: A comprehensive survey: artificial
bee colony (ABC) algorithm and applications. Artif. Intell. Rev. 42, 21–57 (2014)
7. Hu, Z., Zhao, M.: Simulation on traveling salesman problem (TSP) based on artificial bees
colony algorithm. Trans. Beijing Inst. Technol. 29(11), 978–982 (2009)
8. Wong, L.-P., Low, M.Y.H., Chong, C.S.: Bee colony optimisation with local search for
traveling salesman problem. Int. J. Artif. Intell. Tools 19(03), 305–334 (2010)
9. Zhong, Y., Lin, J., Wang, L., Zhang, H.: Hybrid discrete artificial bee colony algorithm with
threshold acceptance criterion for traveling salesman problem. Inf. Sci. 421, 70–84 (2017)
10. Pezer, D.: Efficiency of tool path optimization using genetic algorithm in relation to the
optimization achieved with the CAM software. Procedia Eng. 149, 374–379 (2016)
11. Karaboga, D., Akay, B.: A comparative study of artificial bee colony algorithm. Appl. Math.
Comput. 214, 108–132 (2009)
12. Talbi, E.-G.: Metaheuristics: from Design to Implementation. Wiley, Hoboken (2009).
University of Lille-CNRS- Inria
13. Pezer, D.: Planning of drilling sequence using the swarm intelligence method. In: 9th
International Scientific Conference Management of Technology - Step to Sustainable
Production, MOTSP2017, Croatian Association for PLM, p. 8 (2017)
Comparison of the NC Data Preparation
Methods for Drawingless Production
Peter Kostal(&), Andrea Mudriková, Daynier R. D. Sobrino,

David Michal, and Simon Lecky
Faculty of Materials Science and Technology,

Institute of Production Technology, Slovak University of Technology,
J. Bottu 25, 91724 Trnava, Slovak Republic
peter.kostal@stuba.sk
Abstract. The aim of this paper was to compare the CNC machining data and
CNC programming by using a CAD/CAM system and a workshop program-
ming system. CAD/CAM System CATIA demonstrates the importance and
relationship of new technologies, materials, machines, progressive methods and
information technologies that enable more efficient use of materials source and
achieve lower production costs. This is achieved by shortening the final
machining time and eliminating the inaccuracies caused by the multiple
clamping of a workpiece.
Keywords: Machining Code generation CAD/CAM systems

Drawingless manufacturing Intelligent manufacturing
Manufacturing system design
1 Introduction
Recent decades have witnessed important advances related to information technology

and manufacturing. Quality and manufacturing productivity have been improved
through the technologies such as multi-axis machining, high-speed machining, and
ultra-precision machining. This led to the significant changes in manufacturing com-
panies previously based on low-cost standardized products; they started applying the
elements of modern manufacturing, to provide responsiveness in a context of flexible
demand [1].
All new technologies, materials, machinery, progressive methods and information
tools enable more efficient use of materials, a more cost effective production of
reliable and quality products succeeding in the market in a challenging competitive
environment [2].
In an increasingly globalized market, the growing range of products offered at
competitive prices makes companies look for the ways to reduce both the development
costs of their products and the time to launching them. This has made the economic
priorities of manufacturing undergo significant changes: instead of the low cost stan-
dardized products, companies started making use of the facilities of modern manufac-
turing within the concept of flexible demand, in which a wide range of parts of varied
and complex geometries are produced [1]. As a result of globalization, companies’

https://doi.org/10.1007/978-3-030-18715-6_6
Comparison of the NC Data Preparation Methods for Drawingless Production 63
technological information is becoming increasingly distributed, and it is therefore

necessary for modern manufacturing systems to be robust, intelligent and distributed.
The technological system CNC machine – cutting tool – workpiece – fixture must
ensure sufficient stiffness, which has a direct influence on the quality of the surface
finish of the final product. This technological condition is important to enter the process
of manufacturing of various precision engineering components of complex shape
designed for the needs of industries. Using the highest level of programming through
graphical engineering system is designed for generating the ISO programs for CNC
machine tools [2].
Choice of a CNC program generation as well as the CNC program itself may
ultimately affect the resulting production efficiency. A complex and extensive CNC
program is hard to read and difficult to edit manually. There is a high probability of
error occurrence. Short and simple CNC programs are therefore those based on
machine cycles. However, such CNC programs are usually not sufficiently transferable
between different machines (different control systems).
When choosing the method of CNC programming, it is also necessary to consider
both, the complexity and time-capacity of the design as well as the complexity of the
component to be manufactured.
The article presents two methods of generating a CNC program. Method 1 uses
CAD/CAM modules in CATIA, where a CNC program is generated based on 3-D
model components. Method 2 uses a simple WinNC programming tool. In both cases,
the same parts (manufacturing of which the CNC program was generated for) were
used. The generated programs were then compared to each other.
2 Flexible Manufacturing System
Flexible Manufacturing Systems provide a fast response to the changes in production

conditions. To achieve a quick response of production, it is necessary to abandon the
traditional form of production process planning.
Flexible manufacturing system is a system able to respond to changed conditions.
In general, this flexibility is divided into two key categories and several subcategories.
The first category is the so called machine flexibility which enables to make various
products by the given machinery. The second category is routing flexibility enabling to
execute the same operation by various machines. Flexible manufacturing systems
usually consist of three main parts: CNC machine tools, transport system and control
system. A higher level of flexible manufacturing systems is represented by the so called
intelligent manufacturing systems [3].
In the Institute of Production Technologies, we built a laboratory of flexible pro-
duction systems for drawingless environment. (Fig. 1.) The aim was to build up a
laboratory with a flexible manufacturing system consisting of two NC machines
(milling machine and lathe) where the machine tools are interconnected by a transport
system and operated by industrial robots. The flexible manufacturing system also
comprises a quality control station including camera system and shelf storage [4].
Traditional production systems (inflexible manufacturing systems) are currently
being used for large-scale and mass production in particular. They are used to produce
64 P. Kostal et al.
1 – Main control PC, 2 – conveyor, 3 – lathe, 4 – mill, 5 – storage, 6 – assembly, 7 – testing
Fig. 1. Flexible manufacturing system iCIM3000 [4]
only a very limited range of products. However, the production is highly efficient and
production costs per product are relatively low. The structure, material and information
flow in the systems are steady and usually do not change during production.
Change of the production portfolio in such systems is usually accompanied by a
change in the physical structure of the production system, which makes the changes
costly. Such systems are currently used mainly in large-volume and mass production of
individual components or simple products (screws, bearings, etc.).
The products are commonly available on the market in large quantities and their
purchase can be planned by the customer with sufficient precision far in advance.
Flexible manufacturing systems can typically produce a wide variety of different
products without the necessity to interfere into the actual physical structure of the
system. Both, the material and information flows in flexible production systems are not
stable and can be dynamically modified according to the needs of the current
production.
Change of the production program of a flexible production system generally
requires reprogramming of individual devices. Such changes are quick, relatively
simple and inexpensive. However, production itself is usually less efficient, and thus
production cost per product is higher. This is usually caused by the need for higher
versatility of the system which cannot be highly optimized for a given task.
Flexible production systems are used in the production of smaller series of many
different products, including those of complex surfaces. They usually involve either the
products the customer does not keep in stock, or the ones the delivery of which cannot
be planned far in advance.
In such a case, the time to response to customer requirements is particularly

important. The customers frequently find they need only a few pieces of a complex
product virtually instantaneously, or they may never need it anymore.
The main advantage of a flexible manufacturing system is its high flexibility in
management of production facilities and resources (time, machines and their utilization,
etc.). The widest application of the systems is in the area of small batch production
where its efficiency is getting near to that of the mass production. Its disadvantage is
high implementation price.
This system and some of its parts are in detail described by other papers like [5–12]
and a lot of others.
One approach to the improvement of an engineering design is through reusing
previous knowledge. Organizations in mature markets are in a special position to
benefit from knowledge reuse for three key reasons:
• they know the product well, so are able to produce high-quality reusable
knowledge;
• the next generation product is likely to have a significant overlap with the previous
version;
• the knowledge reuse allows more time for innovation, which is especially important
since competitive advantage is difficult to achieve in mature domains.
Development time, product quality and customer value are all factors which
effective requirements management can improve [13].
The development and implementation of new products through contemporary
production methods and processes is a key issue for achieving international competi-
tiveness. Progress in new technologies such as the various processes of additive pro-
duction, high-speed machining and others is fast.
The current trend is a component machining at one setting and clamping. In this
way, we can achieve a reduction in the total machining time, and eliminate inaccuracies
caused by multiple clamping.
The computer graphics introduction allows creating 3D models of products,
manipulating with them and examining them. Utilization of computer-aided systems
gives significant benefits. The CAD/CAM systems are most closely related to the
features in manufacturing engineering, such as product design, process planning, and
numerical control (CNC). CNC machines programming can be done on three levels:
• DIN ISO programming - manual entry into G and M codes. An advantage is precise
programming of each movement. The disadvantage is high elaborateness and the
possibility of errors.
• Workshop Applications for CNC Programming (Workshop Programming) - These
applications require basic knowledge of DIN ISO programming. They usually have
a preprogrammed cycle library. The workpiece information is graphically specified.
The advantage is the possibility of simulation. The disadvantage is that it is gen-
erally impossible to load a 3D CAD model. Their own CAD module is usually
limited. Complex components are difficult to model.
• CAD/CAM systems - complex production process preparation. CAD is a computer
aided design for production preparation. Are assigned to design 2D/3D product
66 P. Kostal et al.
models, develop technical documentation and other related functions. CAM pro-
vides a possibility of machining simulation. Integration of CAD/CAM offers the
opportunity to generate the CNC code required for the parts production.
In case of machining various complex shapes in one machined part by milling, hole
drilling, etc., we need several clamps on different machine tools to allow each work
step to be performed. In the automated manufacturing process of such components, it is
possible to perform all operations with a minimum number of clamps on multi-axis
CNC machining centers.
3 Part Model as a Manufacturing Information Holder
Modern manufacturing enterprises comprise the facilities spread all around the globe,
which use equipment from hundreds of different manufacturers. Immense volumes of
product information must be transferred between the various facilities and machines.
Despite the advances in manufacturing technologies, the language that computer
numerical control machines (CNC) use nowadays is the G-code (ISO 6983 standard),
which is a set of instructions containing the movements that the machine has to per-
form, without any semantic content regarding the part being processed, causing the loss
of much information about the product and making it difficult or impossible to share
such information across different areas of the company. In contrast, the ISO 14649
standard (known as STEP-NC) provides a high-level data model related to the man-
ufacturing process, and it has been considered adequate to overcome the problems
mentioned above [1].
Today’s digital communications standards have solved the problem of reliably
transferring information across the global networks. For mechanical parts, the
description of product data has been standardized by ISO10303 (STEP). This enables
utilization of standard data throughout the entire process chain in the manufacturing
enterprise. Barriers to implementation of this principle are the data formats used on the
machine level. Most computer numerical control (CNC) machines are programmed in
the ISO 6983 G-code language. Programs are typically generated by computer-aided
manufacturing (CAM) systems that use computer-aided design (CAD) information
[14].
Machining using the STEP-NC standard is called “Worksteps”. “Worksteps” splits
each machining operation into the simple steps necessary to perform the whole oper-
ation, and can be considered as similar to the traditional cycle in the traditional G code.
The STEP-NC standard provides a library of operations that can be performed on a
CNC machine.
In the concept of work steps, the manufacturing processes are simplified. The
machine receives a STEP-NC data, understands its meaning, and manufactures a
component without any further instructions.
Characteristics of the STEP-NC standard give rise to a considerable effort in the use
of STEP-NC in production. The component is inserted into the system into the STEP
format, and the system performs the function recognition that is responsible for linking
the work steps (operations) with the functions, and the work steps are arranged in the
process sequence chart that contains the alternatives. The system deals with intersecting
elements and the part is manufactured by a machine compatible with the STEP-NC
standard.
3.1 Simulation of CNC Machining

Nowadays, most of the products are designed by using CAx software [4]. Use of
CAD/CAM systems makes the entire production process faster and is especially
important for complex-shaped components. In addition, CAM systems provide access
to a wide range of technological and non-technological data such as feed rate, cutting
speed or tool length, number of tool teeth etc.
Individual tasks required machining in the CNC machining centers in the
CAD/CAM CATIA V5 Prismatic Machining modules are as follows:
• Create a 3D model of a component. Correct determination of the geometric
BODY SET for the component.
• Create a raw part model with add-ons.
• In the CAD/CAM module, set the parameters of the tools that will machine the
model.
• Define the strategy of machining operations that are determined by a partial model,
machining permissiveness, tolerance, cutting speed, and stroke of the machine tool.
• Define the machine. Preparation of machine tools for simulation of machining.
• CNC program in ISO format, as appropriate CNC machine tool and CNC control
system.
Methodological approach for the NC program generation by CATIA is described,
for example, in [15]. Figure 2 shows an example of the preparation of the parts
machined in CATIA.
Fig. 2. A part preparation for machining in CATIA

68 P. Kostal et al.
After generating a technological process of production and choice of an appropriate

manufacturing strategy of roughing and finishing of the functional areas of the drilling
and milling tools, it is necessary to select suitable material, geometry and cutting insert
holder. Then determine the optimum cutting parameters for the roughing and finishing
machining operations (feed, cutting speed).
The generated model and tool along with the raw product are put into the envi-
ronment module of CATIA V5 Manufacturing in the work of the tree
PRODUCT LIST. Design of the production process is at the top of the tree under the
name PROCESS LIST. Process through the menu is listed in NC Manufacturing
module design type CNC machine tool. Then we determine the work origin “W” of the
workpiece coordinate system of the machine and the type of post-processors to gen-
erate a CNC code. Use the function “Auxiliary Operations” to generate the optimal tool
shape and cutting material, tool holders and cutting inserts, and determine the strategy
of the whole production process using machining milling operations. The function
“Tool Path Replay” simulates the process of machining. The function “Generating NC
code” then generates the CNC program in ISO format (G and M codes), transferred to a
control system for the CNC machine tool. The choice of a machining strategy for
machining tool path generation has two main objectives: to avoid significant feed rate
loss and to reduce machining time as possible. To achieve these goals, the machining
assistant often associates to a standard of 3-axes machining two types of machining
strategies widely used in CAM software for the machining of forging dies:
• parallel planes strategy: the machining feed direction is defined by parallel planes
containing the tool axis,
• Z-level strategy: the machining feed direction is defined in parallel planes per-
pendicular to the tool axis.
We created several models of several simple parts using different methods. For the
selected components, the corresponding NC codes were generated and the machining
simulation was performed. We tracked the produced NC programs. The WinNC and
CATIA software was used to model and generate NC programs, both of which were
equipped with a post processor for the Sinumeric 840D milling system. One of the
simulated components is shown in Fig. 3.
In simulating the part machining, the planar milling, pocketing operations as well as
the operations for drilled holes were used.
To generate a NC program, it is necessary to define the basic conditions, shape and
dimensions of the component and evaluate the possibilities of clamping the component
into the machine. The next step is defining the tools used and setting up a 3D simulator
so that the machining process can simulated how the component is manufactured. The
settings must also be defined when NC code is imported from the external CAx
software. After generating the NC code, the accuracy of the generated NC programs
was verified by simulation. After the design and simulation of machining, a control
program for the selected machine was generated.
It was agreed that we would work with the same component, and the same tools
available on the CNC machine should be used. However, procedures vary. The dif-
ference was caused by the software running on different NC programming principles.
Fig. 3. One of simple part 3D model
WinNC tries to use components to create 3D elements from a 3D model, and

recognizes individual elements by a predefined cycle from its library, and therefore, at
first glance, the NC code looks easier. As a result of the way the NC program is built,
there are many called functions and unconditional jump in the code that ultimately
make reading and checking more difficult. In such a case, we must make a 3D model of
part by unfriendly environment of a built-in modeler. This kind of a part modeling was
time consumptive.
CATIA V5 generated the tool movements on the base of a 3D model. The NC
program is relatively large, but straight and without an external cycle calling. The part
of modeling is very comfortable in CATIA and designing very complex parts.
Defining the basic machine parameters, tools, and so on also was significantly
easier and more intuitive in CATIA (Fig. 4).
Fig. 4. Tool definition dialog WinNC vs. Catia

70 P. Kostal et al.
All the operations, namely the plane milling, were performed on the same
part. (Fig. 5) In the case of more complex milling operations (contouring, pocketing),
the difference in the NC code is even greater.
Fig. 5. NC code and machining simulation CATIA vs. WinNC on one of the machined parts
For the parts of complex geometry, the use of CATIA V5 software is also more
suitable because it is also a CAD of higher level, where it is easier to model a com-
ponent or part of a free-form shape, but it is technologically possible to work with one
tool, while in CATIA V5, all of these shape faces and parameters for these geometries
can be included in one program.
4 Conclusion
All of our machining simulations show that the tool path proposed works well in the
real machining process and can improve the machining efficiency. At the same time,
the manufacturing process must have modern processing equipment and a method due
to the requirements for performance and machining efficiency.
The generated NC codes in both kinds of software take different forms. By com-
parison, we found out what was common and what was different with these practices.
Both variants of the basic procedures are suitable for the manufacture of the components.
The results presented in this document can be further utilized in the course of CNC
machine training and programming and engineering practice. In our future research
plans is comparison the achieved results in other existing CNC machining centers.
The STEP-NC standard provides an object-oriented data model approach, which
helps to integration. STEP is introduced as a way of transferring data between different
CAD systems, but STEP has also been developed towards to the production infor-
mation management. STEP-NC realize a complete description of the component,
manufacturing process, as well as annotation of CAD design data with manufacturing
information, cutting parameters and tool requirements. STEP-NC defines data repre-
senting work steps, a library of specific machining operations realized on the machine.
STEP-NC is a new data transfer model between CAD/CAM systems and CNC
machines that can replaces the G code.
Acknowledgment. This paper was written thanks to the national KEGA Grant: 021STU-4/2018
- Development of a laboratory for the design and maintenance of production systems supported
by the use of Virtual Reality.
References
1. Ferreira, J.C.E., Benavente, J.C.T., Inoue, P.H.S.: A web-based CAD/CAPP/CAM system
compliant with the STEP-NC standard to manufacture parts with general surfaces. J. Braz.
Soc. Mech. Sci. Eng. 39(1), 155–176 (2017)
2. Dubovska, R., Jambor, J., Majerik, J.: Implementation of CAD/CAM System CATIA V5 in
Simulation of CNC machining process. Procedia Eng. 69, 638–645 (2014)
3. Koštál, P., Mudriková, A.: Laboratory for drawinglesss manufacturing, rput, vol. 26, no. 42,
p. 145 (2018)
4. Košťál, P., Sobrino, D.R.D., Holubek, R., Ružarovský, R.: Laboratory of flexible
manufacturing system for drawinglesss manufacturing. In: Applied Mechanics and
Materials, vol. 693, pp. 3–8, December 2014
5. Danišová, N.: Digital image processing in the camera system of assembly systems ICIM. In:
Applied Mechanics and Materials, vol. 474, pp. 173–178, January 2014
6. Delgado, D.R., Košťál, S.P., Vavruška, J.: On the analysis and customization of an iCIM
3000 system: a take on the material flow, its complexity and a few general issues to improve.
In: Applied Mechanics and Materials, vol. 474, pp. 42–48, January 2014
7. Holubek, R., Ružarovskỳ, R.: The methods for increasing of the efficiency in the intelligent
assembly cell. In: Applied Mechanics and Materials, vol. 470, pp. 729–732 (2014)
8. Holubek, R., Sobrino, D.R.D., Ruzarovsky, R.: Analysis of the communication methods of
an iCIM 3000 system within the frame of research purpose, analysis, vol. 1, p. 5942 (2014)
9. Košťál, P., Krajčová, K., Ružarovský, R.: Material flow description in flexible manufac-
turing. ALS 4, 104–108 (2010)
10. Hrušková, E.: Relative term of capacity computations and manufacturing system design. In:
Applied Mechanics and Materials, vol. 474, pp. 55–60, January 2014
11. Charbulova, M., Matusova, M., Caganova, D.: Intelligent production systems and clamping
systems for intelligent producion systems. J. Prod. Eng. 14(1), 63–66 (2011)
12. Bučányová, M., Riečičiarová, E.: Specification of the component base for CNC processing
centre EMCO Concept MILL 105. In: Applied Mechanics and Materials, vol. 693, pp. 9–15
(2014)
13. Baxter, D., et al.: A framework to integrate design knowledge reuse and requirements
management in engineering design. Robot. Comput.-Integr. Manuf. 24(4), 585–593 (2008)
14. Tóth, D., Mudriková, A., Košt’ál, P.: The role of step-nc in drawingless environment. Acta
Technica Corviniensis-Bull. Eng. 10(3) (2017)
15. Tóth, D., Košťál, P.: The methodic procedure for creation of nc programms by using
Sinumerik 840 D. Acad. J. Manuf. Eng. 12(4) (2014)
Characteristics of Resources as a Determinant
of Implementation of the Physical Internet
Concept in Supply Chains
Michał Adamczak1(&), Łukasz Hadaś2, Agnieszka Stachowiak2,

Roman Domański2, and Piotr Cyplik1
1
Chair of Logistics Systems, Poznan School of Logistics, Poznan, Poland
michal.adamczak@wsl.com.pl
2
Faculty of Engineering Management, Poznan University of Technology,
Poznan, Poland
Abstract. Contemporary companies strive for increasing efficiency of logistics

processes. Efficient logistics processes allow for implementation of corporate
goals with reduced use of resources. Considering the evolution of the concept of
resources, authors in the article focused on the concept of Physical Internet (PI).
For the study purposes, authors of the paper have limited the Physical Internet
idea to sharing resources in supply chain. In the article, results of two survey
studies are presented. The first was conducted among academic teachers spe-
cializing in management and logistics area. The second study was conducted
among managers of production and logistic companies in Poland’s territory. The
first survey study aim was to identify importance of sharing resources as a way
to increase efficiency of supply chains and to identify key determinants of
developing PI concept at resources level. The main aim of the second study was
to get the managers’ opinion on whether it is possible to share resources in
supply chains. Potential difficulties in sharing resources with respect to their
division into material and non-material ones and possible supply chain func-
tioning areas were identified.
Keywords: Sharing resources Survey study Supply chain integration
1 Introduction
The resource use efficiency is of key importance in contemporary conditions of the

world economy oriented to fast meeting various customers’ needs [1, 2]. In the con-
temporary economy, resources should be considered not as a cost generator but a
developer of potential opportunities. A company (and entire supply chains as well) is
given an opportunity to obtain a competitive advantage by benefiting from various
(external) resources and by efficiently using its own resources.
Much attention in the reference literature is paid to supply chain integration [3–6],
which is manifestation of effective sharing of resources. New integration models,
integration level assessment models and supply chain maturity models are established.
The above mentioned solutions are most frequently related to the process approach to
supply chains management. The main aim of the paper is to identify influence of

https://doi.org/10.1007/978-3-030-18715-6_7
Characteristics of Resources as a Determinant of Implementation 73
characteristics of resources on possibility of implementation of Physical Internet

concept into supply chains. Identification was based on two steps survey studies
conducted within academic environment (during ICPR 2017 conference) and among
managers from production and logistic companies located in Poland.
2 Theoretical Background
2.1 Physical Internet
The Physical Internet is a new proceeding concept in the logistics area. The Physical
Internet concept is based on the physical mobility of logistic resources. According to
the Physical Internet concept, all supply chain partners – manufacturers, providers of
transport services, retailers – will be able to function independently by using one shared
logistic network. Its natural feature is the ability to make self-adaptations caused by
turbulent environment [7]. Efficient use of resources (supporting transport and ware-
house processes realization) is the challenge for contemporary logistic systems. The
Physical Internet concept is striving to use better the resources that are currently used
inefficiently, hence it has the potential to be the future option of improving the activity
efficiency in supply chains.
The Physical Internet is the term that was first mentioned in 2006 by Benoit
Montreuil from Université Laval in Canada. The article entitled ‘The physical Internet.
A survey of logistics’ published in ‘The Economist’ includes the first presentation of
the Physical Internet assumptions [8]. The Physical Internet concept is based on the
World Wide Web structure. As information might be transmitted by the World Wide
Web network, why should one not do the same with goods that might be sent by means
of the global logistic network? This requires close cooperation of companies involved
(based on process integration and resources sharing) and results in increase of effi-
ciency of global goods flows – flexibility and performance improvement combined
with the reduction of operational costs. The supply chain configuration, business
models and value adding patterns are being redefined by the Physical Internet
assumptions. This is because the need for searching for a new system solution is more
and more noticeable. The system solution is to enable the increase in the process
performance efficiency and logistics development with the simultaneous obtainment of
economic, social and environmental balance [9].
The ‘Physical Internet’ term was first mentioned in Poland during the Polish
Logistics Congress LOGISTICS 2012 as a part of the paper by Russell D. Meller from
the University of Arkansas [9]. Professor Meller presented the results of simulation
research. It indicated that it was possible for the supply chain to obtain considerable
benefits by shortening supply cycles with a decrease in the negative impact on the
environment.
Inspired by the paper, the team of authors conducted research on worldwide lit-
erature resources that has led to identification of two periods of enhanced publicising
activities in the Physical Internet area [Source: Web of Science]:
• the first period related to the Physical Internet term and idea development (2006 till
2012),
74 M. Adamczak et al.
• the second period of time related to the organizing international conferences on the
Physical Internet topic (2013 and consecutive years).
The PI issues are divided into two approaches:
• technical-technological approach – focused on problems of unification and inte-
gration of logistic units in the supply chain and the infrastructure to facilitate the
flow of these units;
• organisational approach – related to developing the concept of managing the flow of
logistic units which is predominantly based on the possibilities to share resources
and competences with other supply chain participants.
To sum up, the Physical Internet is a young concept that has been actively dis-
cussed in theory and from practical perspective for the last 4 years (since 2012 –
worldwide, since 2014 – in Poland). However, this concept is dynamically developing.
This is confirmed by the small number of conferences and scientific articles and
moderately small number of implementation projects to be currently considered as pilot
solutions [10].
2.2 Resources in Supply Chains

Resources according to the APICS - American Production and Inventory Control
Society [11] dictionary is all that is needed to manufacture the product, and the lack of
which would cause failure to implement the predefined production plan. Hence, the
resources are raw materials and materials (both basic and auxiliary), the potential
calculated as the time of availability of the machines and employees with appropriate
competences, power supplies and money necessary to run manufacturing processes.
The issue of enterprise resources is inseparably connected with the theory of
management and economics and has already appeared in the works of classical
economists. The resources originally described as factors of production were discussed
by: Petty [12], Smith [13], Ricardo [14], Say [15], Marks [16], Marshall [17], Clark
[18], Mill [19], MacCulloch [20] who were pointing to the importance of work,
qualifications, tools and capital for the economy as a whole, and for individual eco-
nomic processes.
The basis of the contemporary resource approach was developed by Schumpeter
and Penrose. Schumpeter [21] considered innovations introduced by entrepreneurs as
crucial for achieving profit, development and economic growth. He defined innovations
as new combinations of material elements and manpower, which contribute to devel-
oping new product or launching a product with new properties, using new production
method, finding new market, acquiring new sources of raw materials or introducing
new production organization. The ability to take innovative actions results in achieving
competitive advantage. Penrose [22] in turn, stressed that the company is a unique set
of production resources that can be used in a variety of ways, and this diversity
translates into uniqueness and, consequently, the ability to gain a competitive
advantage.
The approach based on the assumptions described above is referred to as Resource
Based View of the Firm (RBV). A milestone for the development of the resource
approach was the research by Wernerfelt [23] and Rumelt [24]. The resource theory
presents the enterprise as a unique bundle of material and non-material resources and
skills, which are a source of competitive advantage [25–28], while strategies expressed
in the resource language based on resource redundancy are indicated as the most
flexible and therefore suitable in contemporary conditions. The resources most valuable
for the company are those that enable achieving a sustainable competitive advantage
and organizational success, namely those that meet the so-called VRIN condition, (i.e.
characterized by a strategic value (V-Valuable), rarity (R-Rare), inimitability (I-
Inimitable) and lack of substitutes (N-Non-substitutable)) [27]. The other characteris-
tics referred to in the literature on the subject include rarity, low mobility, limited
opportunity of imitation, substitution, possibility of appropriation, durability, mutual
complementation, adjustment to strategic factors of the industry [28], as well as
imperfect imitation, imperfect substitution, imperfect mobility, diversity, ex ante and ex
post restrictions [29], the concepts of other authors have broadened and integrated the
presented lists [30]. Such configuration of resources should be the basis for the for-
mulation of strategies.
Apart from identifying the essential features of resources as part of the resource
orientation, their classification is also made so that for the identified groups of resources
with common characteristics, appropriate methods and management tools could be
selected. One of the most frequently implemented ways of classifying resources is the
division into material and non-material ones [31, 32]. Classification of resources into
material and non-material is based mostly on the resources existence - material
resources such have actual physical existence, whereas non-material resources exist in
abstraction [33].
Considering the VRIN criteria, researchers suggest that non-material resources are
the source of competitive advantage as they are not easily acquired and replicated in
factor markets [34–36]. The nature of non-material resources includes two aspects:
assets, such as intellectual property, contracts and databased, and skills [37], which
makes them crucial for company’s identity, culture, uniqueness and therefore com-
petitive advantage. Nevertheless, without integration with material resources and
implementing continuous improvement to make the resources structure dynamic [38,
39], single non-material resources will not built competitive advantage by themselves
[35, 40]. Integration of material and non-material resources can have an influence of
efficiency of processes. Efficiency of the resource usage and running processes is one of
the crucial aspect of developing sustainable competitive advantage of companies [41,
42] This is the reason why material and non-material resources contribution to sus-
tainable competitive advantage should be assessed in the same analysis.
3 Survey Studies
3.1 Framework of the Study
The Physical Internet concept is still considered as the developing idea. After analysis
of publications on the subject, it can be concluded that the concept is now more
dynamically developing in the scientific and technical spheres than in the practical one.
This observation served as the main guideline for developing the research framework –
the study on the resources sharing and its importance as a tool for implementing the PI
concept in supply chains. Due to the academic nature of the concept (implementation is
subject to projects and still at development stage), in first stage of survey study was
conducted among academics. The aim of the study was to identify the significance of
particular characteristics of material and non-material resources for implementing the
PI concept in the supply chains. The aim of the second stage of the study (addressed to
managers of manufacturing companies) was to examine the difficulties and threats of
sharing particular types of material and non-material resources in the supply chains as
they could in become the main inhibitors of implementation of PI concept. The
framework research methodology is divided into parts:
• Stage 1: Survey among academics:
• Developing survey questionnaire
• Survey:
• assessment of the validity of research on the impact of PI concept on con-
temporary supply chains in the aspect of their integration with logistic
processes
• identification of the significance of the determinants of implementing the
concept of the Physical Internet.
• Interpretation of the results, In-depth interview (IDI)
• Stage 2: Survey within Practitioners:
• Development of a questionnaire for managers (practitioners)
• Survey:
• study on urgency and influence on the supply chain functioning by its
selected aspects
• opinions on threats implied by sharing material resources and non-material
resources
• degree of difficulty in sharing non-material resources
• Interpretation of the results
• Final conclusions, further research
In-depth study conducted after the academia survey stressed the important role of
planning and coordinating the use of resources as a key factor for the possibility of
efficient sharing of resources in supply chains. Thus, this factor could play a key role in
the implementation of PI concept. It was decided that in the second stage of study the
planning and integration of plans in the supply chains will be included. In the subse-
quent subsections for each of the stages, the research methodology and their results are
described in detail.
3.2 Stage 1: Academic Survey Study

3.2.1 Step 1: Survey Methodology
According to the developed research methodology, the first stage of the research was
addressed to academics. They were asked to complete the questionnaire on the PI
concept and its implementation potential. The questionnaire was distributed in paper
and electronic version among scientists from all over the world during an international
conference and for the academics not attending the conference for a moth after the
event. The result of the survey were reviewed, and the opinions of 22 academics who
self-assessed themselves as experts in the field were qualified as meeting the research
criteria.
The main objective of the study was: Identification of the significance of deter-
minants of the Physical Internet concept implementation in the contemporary supply
chains. Moreover, two research questions were formulated:
• What will be the impact of the Physical Internet concept on logistics processes
management within the next 10 years (2018–2028)?
• Which of the determinants categories material resources or non-material resources
have greater impact on implementation of the PI concept in supply chain logistics
management?
For the purposes of the study, as a result of brainstorming, 14 generally defined
determinants of the implementation of the concept of the Physical Internet were
selected. The category “others” was added in order to recognize the respondents’
opinions on the completeness of the list.
3.2.2 Step 2: Survey Results

The responses of the surveyed academics to the first question: ‘How do you assess
impact of Physical Internet idea on contemporary supply chains in the aspect of their
integration with logistic processes within the next ten years (2018–2028)?’ proved that
further research in the field can contribute to development of the supply chain man-
agement discipline. In total, 68% of respondents rated the examined impact of the
Physical Internet concept as ‘strong’ or ‘critical’ (see Fig. 1). This means that the
respondents perceive the PI concept as one of the promising and dynamically
60%
50%
40%
30%
20%
10%
0%
not average strong critical
relevant
Fig. 1. Influence of Physical Internet idea on contemporary supply chains in the aspect of their
integration with logistic processes within the next ten years (2018–2028).
developing ideas or one of the crucial ideas that will change operation and strategy of
many companies. In the same only 5% of respondents assessed PI impact as ‘not
relevant’ due to the early stage of the concept development, hence, lacking PI’s strict
definition, its importance is possible but in the distant future.
After validating the research, the next step was assessing the significance of the
determinants of the Physical Internet concept implementation with respect to material
and non-material resources. Respondents were asked to answer the questions acting as
experts advising on implementation of the Physical Internet, and identify internal
determinants of the decision on implementation assessing their importance in 5 points
scale, where 1 is not relevant and 5 extremely important. The answers were analyzed
with the mode of a set of data values, which is the value that appears most often in
answers of respondents. Thus, according to respondents the most important material
resources are (Fig. 2):
Fig. 2. Material resources - the answers distribution of respondents for importance on PI

implementation (in 5 points scale, where 1 is not relevant and 5 extremely important).
The respondents have explained that that their selection was based on the
assumption that companies willing to implement the PI concept need at least basic IT
solutions. Implementation of automatic identification, IT systems (ERP, WMS, etc.)
and Telematic systems (Fleet telematics) and their effective use is an important step
towards the implementation of the PI concept.
In the non-material resources category the respondents indicated that the most
important (according to the mode as a set of data values is the value that appears most
often in answers of respondents) are:
• territorial scope of business activity,

• unique know how and USP (unique selling proposition),
• organizational process maturity.
Identifying Territorial scope of business activity as the most important determinant
from the non-material resources category leads to the conclusion that there is probably
a connection between the territorial scope of the offered logistic services and the need
to cooperate and share resources with external entities. Also, having unique know how
and unique selling proposition together with organizational process maturity were
indicated as strong determinants. Unfortunately, at this stage of the research, it could
not be unequivocally determined whether the influence of these determinants on PI
implantation is positive or negative. Nevertheless, respondents noted that the organi-
zation’s ability to effectively manage processes supporting implementation of strategic
objectives has an impact on the analyzed research problem (Fig. 3).
Fig. 3. Non-material resources - the answers distribution of respondents for importance on PI

implementation (in 5 points scale, where 1 is not relevant and 5 extremely important).
In the ‘others’ category the respondents indicated:

• legislation,
• the actual delivery lead time (to the recipient from each location),
• will to cooperate; susceptibility to creating and maintaining open relationships with
the business environment.
As significant. Identification of the impact of delivery lead time on the decision
concerning implementation of the PI concept is related to the need to meet ever higher
Fig. 4. The comparison of significance for material resources and non-material resources
categories for importance on PI implementation (% of responses for importance in 5 points scale,
where 1 is not relevant and 5 extremely important).
standards of logistics customer service. On the other hand, will to cooperate is a

prerequisite for the need to share resources between market entities determines effective
implementation of the PI concept. Legal regulations may also be a strong determinant,
which may, for example, impose the need for cooperation between entities operating in
a given region or agglomerations.
Comparison of the average assessment for both categories shows a slight advantage
of material resources over non-material resources (Fig. 4).
However, the trend is not strong, which allows to conclude that we are dealing with
a group of strong determinants (from both categories) affecting the analyzed problem.
Differences between the categories of material resources and nonmaterial resources will
probably be revealed at a lower level of analysis, i.e. specific issues for management
practice (implemented partly in the second part of the research addressed to manage-
ment practitioners).
3.3 Stage 2: Survey Within Practitioners

3.3.1 Step 1: Survey Methodology
The results of the first stage of study allowed to determine key characteristics and
relevance of resources affecting the possibility of their sharing in the contemporary
supply chains. In the second stage of the study, the authors wanted to gain knowledge
about the possibility of sharing certain resources in the contemporary supply chains. In
order to customize survey’s questionnaire to the conditions of business practice, it was
necessary to identify specific resources of contemporary supply chains (resources with
the characteristics studied in the first stage of survey). And so in an in-depth study, the
academics pointed to the key material resources of supply chains of manufacturing

enterprises:
• transport,
• warehousing – buildings,
• warehousing – equipment,
• production – buildings,
• production – means of labour,
• R&D.
To each of the resources listed above, characteristics that were considered in the
first stage of the study were assigned. The assignment of resources to particular
characteristics is presented in Table 1.
Table 1. Linking supply chain material resources (MR) with the surveyed characteristics of
resources.
Characteristics of resource Resources in supply chain of manufacturing
companies
MR1: Number and size of resources owned Transport, Warehousing – buildings,
Production – buildings, R&D infrastructure
MR2: Fitness of resources owned Warehousing – equipment, Production -
means of labor
MR3: IT systems implemented (ERP, Warehousing – equipment, Production -
WMS, etc.) means of labor
MR4: Telematic systems implemented Warehousing – equipment, Production -
(resources control) means of labor
MR5: Automatic identification systems Warehousing – equipment, Production -
(GS1, RFID) and data interchange systems means of labor
(EDI) implemented
MR6: Resources layout (allocation and Transport, Warehousing – buildings,
availability) Warehousing – equipment, Production –
buildings, Production - means of labor, R&D
Similarly, for material resources, Academics in in-depth study indicated the most
important non-material resources of contemporary supply chains:
• strategic plans,
• sales forecasts,
• data on customers and contractors,
• resource usage degree,
• R&D.
For non-material resources, specific resources were linked to given characteristics.
The relations are shown in the Table 2.
During the in-depth study, the academics pointed to the planning and coordination
of resource use as a key factor in the possibility of their sharing. Without planning and
Table 2. Linking supply chain non-material resources (NR) with the surveyed characteristics of
resources.
Characteristics of resource Non-material resources in supply chain of
manufacturing companies
NR1: Competitive position Strategic plans
NR2: Range of the offer Sales forecasts
NR3: Territorial scope of business activity Strategic plans, Data of customers and
contractors
NR4: Organizational process maturity Strategic plans, Resource usage degree
NR5: Efficiency at operational level Resource usage degree
NR6: Flexibility of operation and Resource usage degree
innovation transfer
NR7: Unique ‘know how’ and USP (unique Data of customers and contractors, R&D
selling proposition)
NR8: Organizational culture Strategic plans
coordinating using of resources, their practical sharing will not be possible. Taking into
account the opinion of academics, it was decided to include the integration of planning
processes in the contemporary supply chains to the second part of the study. Academics
listed the scope of integration of planning processes:
• planning at the strategic level,
• planning at the tactical and operational level,
• operational processes,
• investment actions in the infrastructure area,
• investment actions in the R&D area.
Following the Tables 1 and 2, it was decided to identify connections between the
scope of integration of planning processes and characteristics of material resources
(MR) and non-material resources (NR). These relations are presented in Table 3.
Table 3. Linking characteristics of supply chain resources with scope of integration of planning
processes.
Scope of integration of Characteristics of material Characteristics of non-
planning processes resources (MR) material resources (NR)
Planning at the strategic MR1, MR6 NR1, NR3, NR4
level
Planning at the tactical and MR2 NR2
operational level
Operational processes MR3, MR4, MR5 NR5, NR6
Investment actions in the MR1, MR3, MR4, MR6 NR1, NR2, NR3, NR5, NR6
infrastructure area
Investment actions in the MR6 NR1, NR7
R&D area
Assigning the characteristics of resources to particular scope of integration of

planning processes allows to check the correctness of selection of integration scope in
the context of ongoing research on sharing resources. Covering all ranges of integration
with resource characteristics shows that these ranges relate to the issues under study.
Detailed results of questionnaire surveys concerning the integration of planning pro-
cesses in the contemporary supply chains and the possibility of sharing resources are
presented in the next sub-chapter.
The performed study aimed at collecting opinions on the possibility to implement
practical solutions based on the Physical Internet concept and its postulated resource
sharing in supply chains. The opinions were given by higher-level managers employed
in production and logistic companies. The managers’ opinions were related to such
aspects as:
• urgency of integrating particular areas of the supply chain functioning and its
significance to (production and logistic) companies;
• expected integration effects within the supply chains;
• possibilities to participate in increasing costs and/or increasing supply chain profits;
• possibility to share both material and non-material resources in the supply chains;
• identification of threats imposed by sharing both material and non-material
resources in the supply chains.
The study was performed with the use of a survey questionnaire in the form of both
paper and electronic sheets. The tool selection solely depended on an respondent’s
preferences. The respondent was given the sheet in advance in order to read the
questions and think of answers. The respondent was also given the opportunity to
contact the research team members in order to be able to clarify all the doubts.
42 representatives of production and logistic companies participated in the study.
According to the study assumptions, the respondents represented the positions of
higher-level managers and executives. Among the respondents, there were i.a. chief
operational officers, heads of contracts, department heads, logistics managers. The
respondents represented large companies (the number of employees should be above
250 according to the adopted criterion). The large companies were not accidently
selected due to their largest (organizational and capital) potential to implement modern
resource coordination solutions. All the companies operated in Poland’s territory but
depended on foreign capital as they were (most frequently) part of an international
concern.
The study was conducted in February and March 2017 in the companies that
operated in Poland’s territory.
The conducted study was a pilot study for the further research. The research was
intended to identify whether it was possible to practically implement the Physical
Internet concept. The research results were to determine whether it was justified to
perform further detailed research and the scope of research was interesting to business
practitioners.
3.3.2 Step 2: Survey Results

All the surveyed managers mentioned that it was necessary to perform broader inte-
gration of supply chains. This is obvious with respect to the observed economic
processes and publications discussing them. Nevertheless, it is still justified to ask a

question about which areas should be integrated with each other and with what tools
this integration process should be performed. The research authors indicated selected
aspects of supply chain to the respondents and asked them questions about two specific
integration features:
• implementation priority – defined with respect to the time category, what selected
aspects of supply chain should be integrated with each other in the first place,
• influences of the aspect integration on the supply chain functioning.
The research results are presented in Figs. 5 and 6. In the figures there is also an
average assessment in points (scale 1–5) where 1 means the highest priority and the
Fig. 5. Priority and influence on the supply chain functioning by its selected aspects [40].
Fig. 6. Integration of selected aspects of supply chain significance [43].

largest influence and the percentage of how often a given aspect was mentioned to be
significant in the supply chain integration process.
It should be stated in accordance with the results in Figs. 5 and 6 that particular
selected aspects of supply chain were very similarly assessed by the survey respon-
dents. The assessment was performed with respect to the integration priority and its
influence on the supply chain functioning. Planning integration at the tactical and
operational level were considered to be the aspect of the highest priority. From the
respondents’ perspective, strategic planning integration has the greatest influence on
the supply chain.
As presented in Fig. 6, the integration significance is defined as a product of the
integration priority and influence on the supply chain. The integration significance is an
aggregated measure that was presented both in the category of assessment in points and
the percentage of mentions. There is also a noticeable correlation between the inte-
gration significance and the percentage of mentioning a given aspect by the respon-
dents. The aspects, which were more frequently mentioned, were assessed as more
significant (lower assessment in points). According to this analysis methodology, the
planning integration at the operational and tactical level, planning at the strategic level
and operational process integration were considered to be the most significant to the
supply chain functioning. An essential element in the operational process integration is
to integrate the equipment used to perform processes, including the cargo units in use.
The cooperation within the planning and operation activity performance function might
be related to the necessity to transfer part of planning tasks to other supply chain
entities. Do the managers consider such a situation to be possible and in what condi-
tions? The responses to the questions are presented in Table 4.
Table 4. Opinions on transferring part of planning actions (at the operational and tactical level)
to other supply chain companies [42].
Solution Percentage of
responses
Transfer to other supply chain companies with capital ties with my 52.4%
employing company
Transfer to other supply chain companies regardless of capital ties 23.8%
Transfer of algorithms and electronic infrastructure (collecting and 42.9%
storing operational data)
None of the above 9.5%
While analyzing the results in Table 4, two aspects should be considered. The first
one is rather obvious and shows that it is possible to transfer part of planning functions
to companies that are capital-related entities. The second one is much more interesting
with respect to the possibility to practically implement the Physical Internet assump-
tions and concept. What is indicated by the second aspect, it is relatively highly
acceptable to transfer the planning functions to algorithms and electronic infrastructure.
From respondents’ perspective, this option is more than 2 times probable than the
transfer of planning functions to other companies with no capital ties.
In the follow-up to the topic of sharing resources, the managers of logistic and
production companies seem to notice much more difficulties in sharing non-material
resources. The detailed results within this topic are presented in Table 5.
Table 5. Opinions of the difficulties in resource sharing [43].

Threats to material resources Percentage of mentions
Yes, as no full freedom in using resources 38.1%
Yes, as a risk of getting resources destroyed 28.6%
Yes, as a necessity to coordinate the usage of resources 52.4%
There is no risk related to sharing material resources 9.5%
To conclude the presently described results, it should be stated that there seem to be
numerous difficulties in the most expected cooperation within the operational and
tactical planning. The latter one is treated as an information process. It seems to be
much easier to cooperate within operational process and sharing material resources
within the processes. In the surveyed managers’ opinion, it is expected by the above
action to reduce operational costs (42.9% of mentions and order processing time
reduction – 38.1% of mentions). In the opinion of 57.1% of the respondents, the
obtained results as reduced operational costs in supply chains should be distributed in
the supply chain among the companies that share their resources.
Nevertheless, the resource sharing is not only potential opportunities but also
threats. It seems to be justified to identify the dangers with respect to each of the
resource types (material, non-material) separately. This is motivated by the identified
differences within the possibility to share the resources according to their types.
Table 6 includes the results of the research on potential threats implied by sharing
material resources.
Table 6. Opinions about threats implied by sharing material resources [43].

Threats to material resources Percentage of mentions
Yes, as no full freedom in using resources 38.1%
Yes, as a risk of getting resources destroyed 28.6%
Yes, as a necessity to coordinate the usage of resources 52.4%
There is no risk related to sharing material resources 9.5%
The observation implied by Table 6 is the coordination in using the resources by

various supply chain entities. This is the largest difficulty in sharing the resources in the
authors’ opinion. Importantly, this difficulty is more frequently mentioned than no full
freedom in using the resources and the possibility to get them destroyed.
When moving on to analyzing the difficulty in sharing non-material resources in
detail, the managers of production and logistic companies were asked questions about
the degree of difficulty in sharing them according to the resource types. The responses
to such questions were analyzed as presented in Fig. 7. The respondents were asked to
state whether it was possible to share a certain type of resources and if so, what degree
of difficulty in such actions was based on the scale 1–5 (1 means the highest difficulty
degree, 5 the lowest difficulty degree).
Fig. 7. Degree of difficulty in sharing material resources [43].
In the surveyed managers’ view, it is most difficult to share R&D infrastructure

material resources. Furthermore, it is possible to share means of labor. This is, how-
ever, a difficult task due to their low mobility. The respondents notice large chances in
sharing transport and warehousing resources and they do not identify large difficulties
in the share. Table 7 includes the opinions on the threats in sharing non-material
resources.
Table 7. Opinions on threats from sharing non-material resources [43].

Threats to non-material resources Percentage of
mentions
Yes, information is a competitive advantage element of the 76.2%
company
Yes, as a risk of losing full control over the information access 28.6%
Yes, as a risk 9.5%
There is no risk related sharing non-material resources 4.8%
What indisputably hampers the share of non-material resources, is that they are a
competitive advantage element of a company. In the authors’ opinion, the risk of losing
full information control is of secondary significance compared to the mere information
value for companies. As part of the thorough assessment of sharing non-material

resources, the managers were asked to mention the possibilities to share them and the
degree of difficulty in their share by analogy to the material resources. In this research
an analogical scale (1–5) was used. The research results are presented in Fig. 8.
Fig. 8. Degree of difficulty in sharing non-material resources [43].
The managers of production and logistic companies mention that there is the largest
possibility to share data about the resource usage forecasts and degree. Such actions are
being executed in supply chains within such solutions as CPFR, VMI, CMI or contract
logistics. The largest difficulties are noticed in sharing strategic plans. It is necessary to
emphasize that the indicated degrees of difficulty are very close to each other.
4 Conclusions
All the surveyed managers were unanimous to state that it is necessary to integrate
companies within supply chains. Independently of the survey questionnaire, they
motivated their opinions by the necessity to fulfil customers’ requirements and meet the
increasing market competition. The managers identify the integration as a tool to make
the supply chain obtain a competitive advantage. These observations are in line with
numerous publications about the integration. Thereby, it is not a scientific novelty. It is
current and essential to specify what areas should be integrated with each other in the
supply chain and how it should be performed. The Physical Internet concept provided a
theoretical solution that was oriented to the share of (mostly material) resources. It is
obvious that it is not possible to integrate material resources without any share of non-
material resources. The performed research results showed that the largest potential of
the Physical Internet concept fulfillment occurred in sharing material (transport and
warehousing) resources used in the supply chain operational processes.
It is feasible to state based on the pilot study results that one should consider two
(technical-technological and organizational) approaches in further practical develop-
ment of the Physical Internet concept. The technical-technological approach develop-
ment is observed by numerous researchers and described in numerous publications.
This approach is additionally supported by the fast development of data exchange and
access technologies by wireless networks and M2M communication (and broader IoT
concept). The authors notice larger challenges in the Physical Internet concept devel-
opment in the organizational approach. As shown by the research results, it is necessary
to define new company management models and adapt the currently existing models
for need of implementation the Physical Internet concept and to convince the managers
of coexisting or cooperating companies in supply chains to the models.
The team of authors is planning to perform the next step of the study with respect to
the size of a sample that will enable statistical result verification. The scope of the study
is going to be spread by in-depth study of PI implementation determinants and its
systematization. Due to the international supply chain specifics, it will be necessary to
expend the scope of the research to other countries outside Poland.
Acknowledgments. This paper has been the result of the study conducted within the grants by
the Ministry of Science and Higher Education entitled ‘Development of production and logistics
systems’ (project No. KSL 2/17) pursued at the Poznan School of Logistics.
References
1. Kolinski, A., Śliwczynski, B.: Evaluation problem and assessment method of warehouse
process efficiency. In: Proceedings of The 15th International Scientific Conference, Business
Logistics In Modern Management, Osijek, pp. 175–188 (2015)
2. Koliński, A., Śliwczyński, B., Golińska-Dawson, P.: Model for estimating the economic
efficiency of the production process. Logforum 2, 129–145 (2016)
3. Hadas, L., Cyplik, P., Adamczak, M.: Dimensions for developing supply chain integration
scenarios. Bus. Logist. Mod. Manag. 15, 225–239 (2015)
4. Adamczak, M., Domański, R., Cyplik, P.: Use of sales and operations planning in small and
medium-sized enterprises. LogForum 9, 11–19 (2013)
5. Adamczak, M., Domański, R., Hadaś, L., Cyplik, P.: The integration between production-
logistics system and its task environment chosen aspects. IFAC Papersonline 49(12), 656–
661 (2016)
6. Cyplik, P., Hadas, L., Adamczak, M., Domański, R., Kupczyk, M., Pruska, Ż.: Measuring
the level of integration in a sustainable supply chain. IFAC Proc. Vol. 47(3), 4465–4470
(2014)
7. Hajdul, M., Nowak, P.: Innovative approach to collaboration in joint organization of
transport processes. Logforum 10(1), 51–60 (2014)
8. Montreuil, B.: The physical internet. a survey of logistics. In: The Economist 2006, pp. 1–
15, June 2006
9. Montreuil, B., Rouges, J.F., Cimon, Y., Paulin, D.: The Physical Internet and Business
Model Innovation. Technol. Innov. Manag. Rev 2, 32–37 (2012)
10. Domański, R., Adamczak, M., Cyplik, P.: Physical Internet (PI): a systematic literature. 14
(1), 7–19 (2018)
11. APICS, Apics Dictionary, 15th edn. (2016)
12. Petty W.: Quantulumcunque Concerning Money 1695

13. Smith A.: An Inquiry Into the Nature and Causes of the Wealth of Nations (1776)
14. Ricardo D.: On the Principles of Political Economy and Taxation (1817)
15. Say, J.B.: A Treatise on Political Economy; or the Production Distribution and Consumption
of Wealth. [Translated from the fourth edition of the French]. Batoche Books, Kitchener
(2001)
16. Marks K.: Capital, Volume I [Das Kapital] (1867)
17. Marshall A.: Principles of Economics (1870)
18. Clark J.B.: The Philosophy of Wealth: Economic Principles Newly Formulated (1886)
19. Mill J.S.: Principles of Political Economy, London (1848)
20. MacCulloch, J.R.: A Discourse on the Rise, Progress, Peculiar Objects and Importance of
Political Economy (1824)
21. Schumpeter, J.A.: The theory of economic development: an inquiry into profits, capital,
credit, interest, and the business cycle, New Jersey, (New Brunswick) (1934)
22. Penrose, E.: The Theory of the Growth of the Firm. Oxford University Press, Oxford (1959)
23. Wernerfelt, B.: A resource-based view of the companies Birger Wernerfelt. Strateg. Manag.
J. 5(2), 171–180 (1984)
24. Rumelt, R.P.: How Much Does Industry Matter? Strateg. Manag. J. 12(3), 167–185 (1991)
25. Diericky, I., Cool, L.: Asset stock accumulation and sustainability of competitive advantage.
Manag. Sci. 35(12), 1504–1511 (1898)
26. Reed, R., DeFillippi, R.: Casual ambiquity, barriers to imitation and sustainable competitive
advantage. Acad. Manag. Rev. 15(5), 88–102 (1990)
27. Barney, J.B.: Firm resources and sustained competitive advantage. J. Manag. 17(1), 99–120
(1991)
28. Amit, R., Schoemaker, P.: Strategic assets and organizational rent. Strateg. Manag. J. 14,
33–46 (1993)
29. Peteraf, M.A.: The cornerstones of competitive advantage: a resource-based view. Strateg.
Manag. J. 14, 179–191 (1993)
30. Dytwald, J.A.: Company Resources. Przegląd Organizacja, 3 (1996)
31. Grant, R.M.: Contemporary Strategy Analysis: Concepts, Techniques, Application. Basil
Blackwell, Cambridge (1991)
32. Collins D., Montgomery C.: Competing on Resource Strategy in the 1990s. Harvard
Business Review, July–August 1995
33. Berry, J.: Tangible Strategies for Intangible Assets. McGraw-Hill, New York (2004)
34. Ambrosini, V., Bowman, C.: What are dynamic capabilities and are they a useful construct
in strategic management? Int. J. Manag. Rev. 11(1), 29–49 (2009)
35. Kor, Y., Mesko, A.: Dynamic managerial capabilities: configuration and orchestration of top
executives’ capabilities and the firm’s dominant logic. Strateg. Manag. J. 34(2), 233–244
(2013)
36. Molloy, J., Barney, J.: Who captures the value created with human capital? A market-based
view. Acad. Manag. Perspect. 29(3), 309–325 (2015)
37. Hall, R.: Startegic analysis of intangible resources. Strateg. Manag. J. 13(2), 135–144 (1992)
38. Weigelt, C.: Leveraging supplier capabilities: the role of locus of capability deployment.
Strateg. Manag. J. 34(1), 1–21 (2013)
39. Wang, C.L., Senaratne, C., Rafiq, M.: Success traps, dynamic capabilities and firm
performance. Br. J. Manag. 26(1), 26–44 (2015)
40. Sirmon, D.G., Hitt, M.A., Ireland, R.D., Gilbert, B.A.: Resource orchestration to create
competitive advantage: breadth, depth, and life cycle effects. J. Manag. 37(5), 1390–1412
(2011)
41. Trojanowska, J., Kolinski, A., Galusik, D., Varela, M.L., Machado, J.: A methodology of
Manufacturing. Lecture Notes in Mechanical Engineering, pp. 23–32. Springer, Cham
(2018)
42. Trojanowska, J., Kolinski, A., Varela, M.L.R., Machado, J.: The use of theory of constraints
to improve production efficiency–industrial practice and research results. DEStech
Transactions on Engineering and Technology Research. In: 24th International Conference
on Production Research (ICPR 2017), pp. 537–542 (2017)
43. Adamczak, M., Domański, R., Cyplik, P.: Sharing logistics resources as a way to physical
Internet - results of survey studies. In: Fertsch, M., Stachowiak, A., Mrugalska, B., Oleśków-
Szłapka, J., Hadaś, Ł., Cyplik, P., Golińska-Dawson, P. (eds.) 24th International Conference
on Production Research (ICPR 2017), pp. 774–779. DEStech Publications Inc., Lancaster
(2017)
Model of a Knowledge Management for System
Integrator(s) of Cyber-Physical Production
Systems (CPPS)
Justyna Patalas-Maliszewska1(&) and Nadine Schlueter2

1
Institute of Computer Science and Production Management,
Faculty of Mechanical Engineering, University of Zielona Gora,
Zielona Góra, Poland
j.patalas@iizp.uz.zgora.pl
2
Research Group Product Safety and Quality Engineering,
Faculty of Mechanical Engineering and Safety Engineering,
University of Wuppertal, Wuppertal, Germany
Abstract. This paper explores the problem of the possibility of integrating the
Knowledge Management System (KMS) and System Integrator(s) in Cyber-
physical Production Systems (CPPS). In the first stage of our approach, CPPS
development network, including relations and the applied Systems Engineering
approach was presented. Our next approach was to integrate the CPPS devel-
opment network with the tacit knowledge management system (TKMS); this
was developed and strictly described. Our proposed approach to TKMS, inte-
grated into the system integrator of CPSS (KMStoCPPS) includes: (1) identifi-
cation of processes within the enterprise network that should be supported by the
knowledge management system and also by the system integrator of CPSS,
(2) identification of experts within a network as the source of knowledge,
(3) acquiring tacit knowledge from experts within a network, (4) the classifi-
cation of tacit knowledge within the network, (5) the knowledge base for the
network. Finally, based on the example, the KMStoCPPS model was explained.
The proposed approach can be a useful for manufacturing enterprises working
within a network and can be also treated as an element in the adaptation of an
enterprise to the concept of Industry 4.0.
Keywords: Cyper-physical production system

Model-based systems engineering Network Knowledge management
1 Introduction
The Cyber-Physical Production System (CPPS) is a production system that relies on the
latest in communication technology. CPPS also relies on foreseeable further devel-
opments in information and communication technologies which are to be integrated
into the latest in manufacturing systems. As such, it is the basis for the 4th. Industrial
Revolution, frequently noted as Industry 4.0 [1]. CPPS is a combination of and a
connection to “cyber-physical systems” and the “Internet of services and things”. These
new concepts and technologies are developed continually by scientists [2] and cause

https://doi.org/10.1007/978-3-030-18715-6_8
Model of a Knowledge Management for System Integrator(s) of CPPS 93
changes and updates regarding common production systems. Thus, CPPS has some
special characteristics that lead to new requirements in the field of modelling and the
analysis of the CPPS itself and also regarding knowledge management; existing con-
ventional modelling methods such as Markov Models and Petri nets [3] can, for
example, be used to model production processes. However, it must now be addressed
as to whether those models are still the best choice, if we need to link those process
models with knowledge management models and procedures.
The Knowledge Management System (KMS) can be treated as the connection
between the process models of activities completed within a company and with their
partners and knowledge management tools [4]. A process-oriented enterprise should be
presented as a process map that takes into account relations with business partners.
Designing a Knowledge Management System can be presented by means of notation,
that is, presented as a standard, due to which it is possible to describe economic reality
unambiguously and in a manner which is understandable both for programmers and for
the management staff of companies. Examples of such notations are: Architecture of
Integrated Information Systems – ARIS, Event-Driven Process Chain – EPC, Toronto
Virtual Enterprise – TOVE, Semantic Object Model – SOM, Unified Modelling
Language – UML and Business Process Model and Notation – BPMN.
BPMN is the most commonly used standard among those listed for the graphic
legibility of this language [5] as well as being a facility for use in representing exe-
cution processes [6].
On the other hand, the following knowledge management (KM) tools [7] can be
distinguished, namely, Blogs, Chat, Collaborative Filtering Brainstorming Configura-
tion Management System, Content Management System Casual, Data Management
Systems Data Mining, Data Visualisation, Data Warehouse, Decision Support Systems,
Document Management System, E-mail Knowledge, ERP Systems, Expert Systems
Knowledge, Learning Management System, Mash-up, P2P Resource Sharing Knowl-
edge, Podcasting/Videocasting, Prediction and Idea Markets, Product Data Manage-
ment Systems, Product Lifecycle Management Systems, Social Data Mining, Social
Media, Syndication Systems, Trust and Reputation Systems, Video/audio Conference,
Wiki. In order to integrate the KMS with SI for CPPS, one cannot simply adopt a given
KM-tool, because the solution supporting knowledge management should support the
defined information needs of companies within the network and should be integrated
with business processes implemented in companies. For the purpose of designing
KMS, the processes within a network, supported by the knowledge management
system should be precisely defined; that knowledge which is key to a given process
should then be strictly determined which, in turn, will lead to a dedicated design
solution supporting knowledge management. The conventional modelling methods
mentioned are Markov models and Petri nets; the methods are EMT, Principal Com-
ponent Analysis - PCA, Fuzzy logic, knowledge mapping, semantic net, classification,
regression trees and the Bayes and Monte Carlo Algorithm; these still support the
classification of knowledge and can be part of KMS.
94 J. Patalas-Maliszewska and N. Schlueter
In this paper, the approach to KMS, integrated into the System Integrator(s) of
Cyber-physical Production Systems – CPPS - (KMStoCPPS) is proposed. The com-
mon elements for creating KMS, as the part of the system integrator of CPSS, are
defined as one standard to be used in the modelling of business processes implemented
within a network; this should be supported by information technology and knowledge
bases within the network, in which useful knowledge, from previous projects, can be
stored and made available to the partners in a network.
We argue that the integration of KMS and CPPS could take the form of the
conceptualisation of knowledge [8] within a network and include relations between the
business partners within a network and the Systems Engineering approaches applied.
2 System Integrator(s) of Cyber-Physical Production Systems

(CPPS)
A principal feature of any Cyber-physical production system is its adaptability; it

adapts automatically to new situations, system goals and requirements. One example is
the “smart factory” so-called, which can adapt its production systems to new product
variants while guaranteeing, at the same time, minimum energy consumption in any
production scenario. This means, that CPPS adapts to new product descriptions.
A product is no longer limited to the capabilities of a production system.
In order to develop and implement such a complex CPPS, the collaboration of
different science disciplines is required, viz., informatics, mechanical and electrical
engineering, system engineering, material science, logistics experts, etc., to name but a
few. Therefore, most providers of CPPS are organised in business networks, including
those different kinds of experts with their know-how, so that the common goal of
providing CPPS can be achieved. Those networks are led by a system integrator, who
is the ‘face’ to the customer. While a single network partner is responsible for a few
components, concepts or processes of the CPPS, the system integrator is the one
handling all requirements, managing the whole CPPS lifecycle and assuring
conformity.
As an example, a system integrator (SI) wants to develop, built and provide
maintenance service for a smart factory manufacturing vacuum-cleaner robots. The
network of system integrators consists of several partners:
– Software company for manufacturing plants (SMP)
– Robotic supplier (RS)
– Intra-Logistics company (IL)
– Software company for business administration (BA)
– Company for Additive Manufacturing Technology (AMT)
– Packaging Company (P)
– Company for image processing (IP)
– Company for automatic control engineering (ACE)
While, in this example, the system integrator as the network manager is required
contact to each partner, not all partners are linked to one another. Nonetheless, it is a
stable network, co-operating with one another over several years; even so, internal
communication is organised and filtered by the system integrator on account of having
to protect the ‘know-how’.
In order to develop, implement and maintain the CPPS, the system integrator has
the following solution philosophy and uses a synthesis approach [9] which designs the
physical components of CPPSs and generates the software from scratch, based on
models of the product and on the production resources available. He also develops
services for the CPPS that are individually adapted to his customers´ needs regarding
maintenance and staff training.
Fig. 1. Example of a CPPS development network including approaches to relations and applied
Systems Engineering
Although the system integrator is using an INCOSE-conform Model-Based System

Engineering (MBSE) approach [10], he still faces several problems that are common to
interdisciplinary development networks. As can be seen in Fig. 1, the network lacks a
common, standardised MBSE approach along the whole CPPS life cycle. This causes
gaps in communication and methodology, misunderstandings regarding terms,
incompatible models and thus increasing costs; neither are milestones met, nor
requirements satisfied.
In order to tackle those problems, the system integrator must find solutions to the
following challenges:
– What does a generic MBSE concept for CPPS resemble..?
– How can knowledge from previous projects be stored and made available for the
relevant network partner and staff while simultaneously ensuring the protection of
know-how..?
The first question can be answered by analysing actual, international MBSE con-
cepts for solving interdisciplinary problems. In 2016, an analysis was carried out to
identify possible MBSE approaches and compare them to each other regarding inter-
disciplinary modelling and procedures, etc. [11]. Generic Systems Engineering
(GSE) was identified as the best possible option [12].
GSE is a model-based Systems Engineering approach which uses a meta-model,
called Demand Compliant Design (DeCoDe), as an information pool for all the specific
models of a discipline. This way, all networks partners can still apply their specific
models to a sub-problem, as long as the meta-model is updated with the latest results.
The GSE model uses a minimum of views (types of element) as well as relation types
while being compatible to all different types of discipline specific models. For a
technical system, as an example, it suggests using just the element types “requirement”,
“function”, “component” and “process”. This way, system theory rules for handling
complexity are applied. Furthermore, GSE requires an iterative, problem-solving
procedure including the setting of goals for modules, their analysis and design,
organised via a project management module. Within these modules, the best possible
method for solving the problem at hand can be chosen by the interdisciplinary teams.
This way, there is no focus on discipline specific procedures but rather, the focus is on
interdisciplinary, team-oriented decision-making, regarding co-operation. Also, the
concept of updating the system model at each step of a module is unique. This way, all
team members are provided with an up-to-date system model with which they can work
[12].
But while GSE and its DeCoDe-model has been applied in the design of several
technical, as well as socio-technical systems, (e.g. [13–15]) there is still a lack of
experience in the field of CPPS; nor is there any experience with regard to linking or
enhancing its project management module to knowledge management.
In order to arrive at a solution as to how a knowledge management system can be
integrated into the use of GSE in CPPS, the components of the proposed approach need
to be identified.
3 Approach to KMStoCPPS
Knowledge, in a CPPS development network, may not only be formalised (reports,

catalogues or patents), but may also be tacit, identified by experts, researched and
developed [16–18]. We believe that the need for the creation of an approach to
integration KMS and CPPS (KMStoCPPS) outlines the importance of expert know-how
because it is very important in the development of information systems. The approach
to KMStoCPPS should also be social infrastructure, so-called [19] for the acquisition,
storage, classification and finally the transfer, within a network, of knowledge. In order
to ensure common understanding in a network, as well as create synergies between
knowledge management and project management, we suggest using the same model, in
this case, e-DeCoDe. After identifying all the partners of the CPPS development
network and the manner in which they handle development processes (see Fig. 1), we
enhance the model, according to e-DeCoDe rules, with information on areas of tacit
knowledge; we then need to acquire it and link it to the partners (see Fig. 2).
Fig. 2. Approach to integrating a tacit knowledge management system into CPPS development
network
Our proposed approach to KMStoCPPS includes: (1) identification of processes

within the network which should be supported by the knowledge management system
and also by the system integrator of CPSS, (2) identification of experts within a
network as the source of knowledge, (3) acquiring tacit knowledge from experts within
a network, (4) the classification of tacit knowledge within the network, (5) the
knowledge base for the network (Fig. 3).
Fig. 3. Approach to KMStoCPPS
The five steps of this approach are explained in the following:

(1) identification of processes within the network, that should be supported by the
knowledge management system and also by the system integrator of CPSS:
For each identified company within a network: SMP, RS, IL, BA, AMT, P, IP,
ACE, we should devise a web-based questionnaire for the employees; this will facilitate
realisation of the business processes defined. In a given month, a worker selects the
business process and devotes his/her time, over a set range of hours, to this process, say
1 h–160 h. Then the IT system analysts define the common business processes for the
network, using BPMN notation. The group of workers who will undertake and com-
plete processes within a network, can be defined using the ‘K-means clustering and
distance Method’ algorithm.
(2) identification of experts within a network, as a source of knowledge
For each worker in a company, we can obtain the value of his/her personnel
usefulness function [20, 21] which may be defined as:
Fn ¼ GK þ PK þ A þ E þ CI
where:
Fn - personnel usefulness function
GK - the general knowledge of the worker within a network
PK - the professional knowledge of the worker within a network
A - the professional abilities of the worker within a network

E - the experience of the worker within a network
CI - the capacity of the worker within a network
The values of the personnel usefulness function and of the “acceptance” module for
each worker within a network will be obtained from workers and managers through
interviews/tests conducted at each component of this function. Each worker completes
a questionnaire. Using an algorithm to test solutions for each worker, it is possible to
determine the value for the personnel usefulness function of the worker. The “per-
sonnel usefulness function” for each worker is as a sum of 5 values, because all
variables are equally important for experts’ sources of tacit knowledge within a net-
work. A detailed description of this function and its implementation in the form of a
web application is presented in the work [22].
(3) Acquiring tacit knowledge from experts within a network
The knowledge of identified experts, within a network, can be acquired using the
methods: real-time teacher observations, analyses of problem-solving procedures,
training sessions, courses, demonstrations, auditing knowledge and hidden interviews
[23].
(4) Formal representations of tacit knowledge and transferring it into a computer-
implementable format
The main procedural problem in acquiring tacit knowledge is formally representing
it and transferring it into a computer-implementable format. We can distinguish the
following methods for representing knowledge; this is useful, for the knowledge
conversion process and builds tacit knowledge bases within a network:
• Methods based on the application of logic.
• Methods based on concept dictionaries (ontologies).
• Methods based on the record of statements.
• Methods based on the establishment of rules.
• Methods based on the concept of framework.
• Semantic networks.
Due to the representation of knowledge from previously completed projects, the use
of a method-based establishment of rules is proposed. Each completed project report,
received from an expert, can be converted according to a defined formula. This is then
stored in the knowledge base for the network.
Consequently, we receive a knowledge base (5) -the knowledge base- concerning
projects previously completed within a network and attributed to business processes.
The format that this model base might take, is described in the following section.
4 Tacit Knowledge Base for the Network
In order to illustrate how the KMStoCPPS model might appear, that is, the format it
might take, we use the same example as above but this time, we show a more detailed
sub-level process of the network (Fig. 4).
Fig. 4. Example of the KMStoCPPS model at the second process level
Based on the given processes at the second process level, we can link identified
experts. Each expert has his own usefulness function Fn, assigned as an attribute. The
sum total of the attributes of all the workers in a process can also be combined and
provided as an attribute of the process. In this manner, those processes which are more
important than other processes, for the management of knowledge, can be identified.
By applying the system engineering principle of different detail levels within a
model, the complexity of tacit knowledge can be managed. By linking each tacit
knowledge aspect to processes and persons, that is, workers, we also create categories
and structures that help us to manage the complexity of the system by using them as
filters.
For a more detailed analysis, it is suggested that more detailed attributions for each
worker be applied which will take into account all the components of the usefulness
function Fn viz., the general knowledge of the worker within a network, the profes-
sional knowledge of the worker within a network, the professional abilities of the
worker within a network, the experience of the worker within a network and the
capacity of the worker within a network.
As the main advantage of the models chosen is a minimum standard of element

types, the model is able to add information of more specific models such as the
knowledge management model which uses BPM-notation, by using attribution. Thus, it
can serve as an information-pool for the network. This shows all important facts in a
common, understandable notation. This way, misunderstandings in communication can
be reduced.
On the other hand, there is also a disadvantage. The model, its rules and its
notations are different from the models that are used for developing web-based software
tools for handling knowledge management. Thus, in order to use the information
gained in project management (PM) via an integrated PM-KM-tool, the meta-model
information in DeCoDe must be transferred into the web-based tool.
This, of course will incur additional time and costs. In future case studies, therefore,
one must analyse if the advantages of the standardised meta-model for improving
communication and information takes precedence over the disadvantage of transferring
work.
5 Concluding Remarks
Production enterprises are faced with challenges including costs reduction, demands for
individualised products, to say nothing of increasing demands for higher levels of
performance in the provision of services. These cause special requirements regarding
production planning, control and monitoring. Such new challenges need to be met with
suitable methods and techniques in the field of CPPS management. Within this context,
developments, such as Industry 4.0 and cyber-physical production systems in tech-
nology, along with such approaches as innovative knowledge management, hold great
potential [24].
This paper focusses on the aspect of ‘Knowledge Management in CPPS in
Development Networks’ in order to identify models and procedures for an effective
approach.
Based on state-of-the-art results in Systems Engineering and Knowledge Man-
agement, we have presented a first conceptual approach, outlining how Knowledge
Management, in CPPS development networks can be applied; however, in order to
design an effective approach, further research and case studies need to be carried out.
For example, further analysis and research needs to be done on the possibilities of
linking knowledge management procedures to project management procedures, in
networks, including possibilities for the protection of knowledge. Also, case studies
should be carried out in order to identify degrees of effectiveness and efficiency.
Last but not least, this new approach is only applicable, if it is supported by
appropriate software.
While there are several knowledge management software tools available on the
market, those which can be adapted for network projects should be ascertained.
References
1. Monostroi, L.: Cyber-physical production systems: roots, expectations and R&D challenges.
CIRP Procedia 17, 9–13 (2014)
Machado, J.: Collaborative manufacturing based on cloud, and on other i4.0 oriented
Eng. Rev. 9(3), 90–99 (2018)
3. Long, F., Zeiler, P., Bertsche, B.: Modelling the production systems in industry 4.0 and their
availability with high-level Petri nets. In: 8th IFAC Conference on Manufacturing
Modelling, Management and Control, Elsevier, vol. 49, no. 12, pp. 145–150 (2016)
4. Centobelli, P., Cerchione, R., Esposito, E.: Knowledge management systems: the hallmark
of SMEs. Knowl. Manag. Res. Pract. 15(2), 294–304 (2017)
5. Lu, R., Sadiq, S.: A survey of comparative business process modeling approaches. In:
Business Information Systems, pp. 82–94. Springer, Heidelberg (2007)
6. Makni, M., Hadj-Alouane, N.B., Tata, S., Yeddes, M.M.: Negotiating deadline constraints in
interorganizational logistic systems: a healthcare case study. In: Proceedings of the
International Workshops and Education Track on BPM. LNBIP, pp. 08–118. Springer,
Heidelberg (2011)
7. Centobelli, P., Cerchione, R., Esposito, E.: Aligning enterprise knowledge and knowledge
management systems to improve efficiency and effectiveness performance: a three-
dimensional fuzzy-based decision support system. Expert Syst. Appl. 91, 107–126 (2018)
8. Guan, J., Liu, N.: Exploitative and exploratory innovations in knowledge network and
collaboration network: a patent analysis in the technological field of nano-energy. Res.
Policy 45, 97–112 (2016)
9. Otto, J., Henning, S., Niggemann, O.: Why cyber-physical production systems need a
descriptive engineering approach – a case study in plug & produce. Procedia Technol. 15,
295–302 (2014)
10. INCOSE: INCOSE Systems Engineering Handbook, 4th edn. Wiley, Hoboken (2015)
11. Schlüter, N.: Generic Systems Engineering für die Entwicklung verlässlicher Unternehmen-
snetzwerke. In: Winzer, P. (ed.) Herausforderungen der Digitalisierung. 1. Auflage, Shaker,
pp. 1–17 (2016)
12. Winzer, P.: Generic system description and problem solving in systems engineering. IEEE
Syst. J. 11(4), 2052–2061 (2015)
13. Bielefeld, O., Schlüter, N., Winzer, P.: Development of a methodical approach for
requirements management in cross-company networks (ReMaiN). In: 3rd International
Conference on Quality Engineering and Management, Barcelona, Spain, pp. 632–650 (2018)
14. Bielefeld, O., Dransfeld, H., Schlüter, N.: Development of a procedure for analysis of failure
chains in complex mechatronic systems to improve sustainability. In: 15th Global
Conference on Sustainable Manufacturing, Procedia Manufacturing, vol. 21, pp. 870–877
(2018)
15. Bielefeld, O., Dransfeld, H., Schlüter, N., Winzer, P.: Development of an innovative
approach for complex, causally determined failure chains. Manag. Prod. Eng. Rev. 8(3), 3–
12 (2017)
16. Jashapara, A.: Knowledge Management. Polskie Wydawnictwo Ekonomiczne (2006)
17. Holste, J.S., Fields, D.: Trust and knowledge sharing and use. J. Knowl. Manag. 4(1), 128–
140 (2010)
18. Song, J., Asakawa, K., Chu, Y.: What determines knowledge sourcing from host locations of
overseas R&D operations? A study of global R&D activities of Japanese multi-nationals.
Res. Policy 40(3), 380–390 (2011)
19. Aalbers, R., Dolfsma, W.: Innovation Networks: Managing the Networked Organization.
Taylor and Francis, Hoboken (2015)
20. Patalas-Maliszewska, J.: Knowledge Worker Management: Value Assessment, Methods,
and Application Tools. Springer, Heidelberg (2013)
21. Patalas-Maliszewska, J., Krebs, I.: Modelling an application for tacit knowledge acquisition
support for an IT company. In: Trends und Communities der Rechtsinformatik. IRIS 2017:
Tagungsband des 20. Internationalen Rechtsinformatik Symposions, pp. 275–282 (2017)
22. Patalas-Maliszewska, J., Krebs, I.: Exemplar functionalities of the intelligent system for
tacit-knowledge-transfer support for an IT company. In: Damaševičius, R., Mikašyte, V.
(eds.) Information and Software Technologies: 23rd International Conference. Lecture
Communications in Computer and Information Science, vol. 756, pp. 25–34 (2017)
23. Patalas-Maliszewska, J., Dudek, A.: A model of a tacit knowledge transformation for the
service department in a manufacturing company: a case study. Found. Manag. Int. J. 8(1),
175–188 (2016)
24. Seitz, K.-F., Nyhuis, P.: Cyper-physical production systems combined with logistic models –
a learning factory concept for an improved production planning and control. CIRP Procedia
32, 92–97 (2015)
Knowledge Management in Open Industrial
Virtual Reality Applications
Filip Górski(&), Paweł Buń, Przemysław Zawadzki,

and Radosław Wichniarek
Chair of Management and Production Engineering,

Poznan University of Technology, Poznań, Poland
filip.gorski@put.poznan.pl
Abstract. This paper presents practical results of a new, open approach to

knowledge formalization and management in Virtual Reality applications for
use in industry, especially in design and training purposes. Idea of applications
with open access to knowledge is presented. Three cases of VR systems are
shown – two for the design and one for training, all implemented in industrial
companies in Poland. The focus is put on concepts of storing and easy editing
knowledge in the applications, in a way to ensure self-sufficiency of applications
in a given company after deployment. Dedicated tools for knowledge edition are
presented, along with selected results of their implementation in scope of
effectiveness of expanding the knowledge base.
Keywords: Virtual reality Design Training Manufacturing
1 Introduction
Modern production companies constantly try to find ways to be more competitive.

Today, clients want to have a share in decision-making about shape, functions and
appearance of their products and want them delivered as quickly as possible. That is
why increasing competitiveness of manufacturing companies nowadays consists
mostly in offering individualized, customized products, delivered in a very short time.
This is known as the mass customization (MC) approach [1].
The new trend of Industry 4.0 focuses on smart factories, Internet of Things and
cloud computing, as well as human-machine integration on every possible level. This
assumes digitization of virtually all aspects of production – a concept known previ-
ously as digital manufacturing or computer integrated manufacturing. It requires to
translate the design and production processes into a virtual space, what is possible
thanks to use of the virtual technologies (recently described as XR) – virtual reality
(VR), augmented reality (AR) and mixed reality (MR).
Virtual reality has been utilized industrially for few decades now. First implemen-
tations of VR in mechanical engineering come from the 90s of XXth century, e.g. in
Boeing company [2]. Since the beginning, two aspects were highly emphasized – use of
VR for aiding the design of new products and virtually assisted training of various
manufacturing and auxiliary processes. There are many successful industrial imple-
mentations of VR in both of these aspects, in companies such as Volkswagen [3, 4],
https://doi.org/10.1007/978-3-030-18715-6_9
Knowledge Management in Open Industrial Virtual Reality Applications 105
MAN [5] or Ford [6]. Efficiency of VR relies mostly on the immersion effect, what is a
human feeling of being present in an artificially created world. The VR simulations are
human-centric, focusing on stimulating the senses, particularly sight, hearing and touch,
in order to persuade one’s brain into believing of being present elsewhere. This aids
imagination, allowing more efficient training or design [7, 8]. Recent revolution in both
software and hardware, such as introduction of cheap VR headsets, e.g. Oculus Rift,
drastically decreased prices of VR systems, allowing its accessibility for even the small
and medium enterprises. Many new companies offer development of VR software
applications, due to increasing market demand. The accompanying technologies of
augmented reality (AR) and mixed reality (MR) also gained considerable impact and a
number of applications.
Requirements of production companies, especially considering the mass cus-
tomization and Industry 4.0 premises, make traditional ways of development of VR (or
AR/MR) software obsolete. Most VR solutions are built without any methodology
whatsoever [9] and the most important hindrance, industry-wise, is closed nature of
such applications. It means that knowledge about a given product or process is hard-
coded into the application and any changes (for example, in product shape or sequence
of a process) is not possible from the viewpoint of a manufacturing company, as it
requires going back to the source code, meaning costly and lengthy intervention back
from the original developer.
This paper presents one of the aspects of a methodology of building open VR
systems and applications – knowledge formalization and management in industrial VR
apps. Idea of open-access VR application is presented, along with two cases of systems
with a library of formalized knowledge, assisted with special, dedicated tools for
knowledge editing from inside the original user – a production company.
2 Methodology
2.1 Idea of Open Industrial VR Applications

Two main purposes of use of VR techniques in a modern production company can be
distinguished as below:
– virtual design and prototyping – creation of virtual models of products and pro-
cesses in a company, especially when introducing new ones, allowing easy con-
figuration and checking of features of particular design in an immersive visual
environment [7, 10],
– virtual training – using the aforementioned virtual models for quick and safe learning
of skills and procedural knowledge by employees of a given company [11, 12].
In general, there is a requirement of VR applications to be maintained throughout
the whole lifecycle of a given product or a process. The maintenance is ideally done
inside the company, without need of hiring external developers. Access to knowledge
about process or product should be open and data should be easily changed without
programming. Currently used approaches in development of VR applications are
derived from development methodology of video games, thus not allowing flexibility
106 F. Górski et al.
and openness on a level required by the industry. Additionally, there are no method-
ologies of building VR (or AR/MR) applications for industry, as mentioned by many
authors [9, 13, 14]. Existing methodologies regarding VR in manufacturing are scarce
or limited [15, 16]. No methodology allows building open VR applications – usually it
means that after coding, any access to product/process knowledge is lost to everyone
but the programmer.
Fig. 1. Traditional (a) and new (b) process of building VR applications for engineering [17]
An innovative approach to this problem is proposed, on the basis of a range of

realized projects and implementations of VR solutions in industrial companies – a
knowledge-based approach. The most important change from the traditional approach
is focusing on recording the product/process knowledge in an application in a formal
way and storing it outside the application, for later easy access both by humans and in-
built algorithms. The key idea is presented in Fig. 1 and was described in earlier
publications [17].
The proposed methodology of building open VR applications – Knowledge Ori-
ented Methodology of Development of Virtual Reality Systems (KOMDEVRS) is
based on Knowledge Engineering and Design Thinking approaches. The main stages of
development are presented in Fig. 2.
Fig. 2. Basic stages of the proposed methodology (KOMDEVRS) [17]
The knowledge acquisition and formalization stages are the key stages, at which a
VR system to gains its unique capability of openness (flexibility). Knowledge acqui-
sition is aimed at obtaining all the necessary knowledge in an unstructured (unordered)
way, in all forms possible (2D, 3D, audio, text, etc.). Knowledge formalization consists
in structuring the knowledge, organizing it and assuming a certain ways of recording
various units of the knowledge, from the viewpoint of a specific application. The
further part of the paper briefly overviews basics of knowledge formalization and
management in open VR applications for industry.
2.2 Knowledge Formalization

In the case of building open VR systems according to the proposed methodology
(Fig. 2), knowledge formalization in a technical sense is a process leading to creation
or development of a basic module of the whole VR system, which is the library (also
known as the knowledge base). It can be also related to creation of a second module,
which is the application for knowledge management. The main module – a visual-
ization module – could be created later, as a third, final module, when all the
knowledge management mechanisms are settled down. However, usually a simple non-
open proof of concept of a working VR application has to be created prior to main
development (at the justification stage, for example to convince the stakeholders),
hence in the chronological sense, the client side application comes to life first. What is
noteworthy here, is that in case of the closed VR applications created in a traditional
manner, there is only one module – the client side module and the knowledge base is
hidden inside its code.
Focusing on the knowledge formalization process, it should be divided into the

following phases:
1. Creating a base structure of the open knowledge base, meaning the folder tree, with
regards to specific forms of knowledge (known from the previous stage – knowl-
edge acquisition) and recorded data. The folder structure should be created with
future expansions in mind and is also a product of decisions on the formalization
level and type (described further).
2. Making decisions on file formats stored in the knowledge base, depending on the
form of the knowledge and available data import/export mechanisms in a selected
VR development software (examples: 2D data – JPG, PNG, 3D data – OBJ, STL,
FBX, STP, metadata – TXT, XML etc.).
3. Creating templates of names of data files, with regards to future scalability of the
knowledge base. Making a decision on free or constrained naming of appropriate
files.
4. Creating a list of file types saved in the knowledge base, along with determination
of their format, function, naming and location.
5. Development of structure of metadata files (plain text or special format, such as
XML) – determination of methods of references to particular data files, definition of
language of logic describing instances of objects in the VR application.
6. Filling the folders with proper data files, in quantity allowing to prepare the full
application, along with validation of procedure of knowledge base editing and
expansion.
7. Manual filling metadata files, converting informal knowledge about the product
and/or the process into description of data allowing the VR application to run its
logic realistically.
8. Creation of algorithms of import, export and processing of knowledge stored in the
base by the visualization module of the future VR application.
Such a base of knowledge, completed with algorithms of import, export and pro-
cessing of knowledge by the VR visualization module can be a basis for building the
proper client side application in a selected VR development software (e.g. Unity or
Unreal engine).
The formalization is not always required to be complete, meaning that not all the
knowledge is stored outside the main VR application. Sometimes part of the knowledge
(particularly 3D objects) can be hard-coded without losing flexibility or at the expense
of time needed for implementing the full formalization scheme. This is particularly true
for 3D models, which are generally difficult to import and process during runtime of
any VR application. That is why a concept of formalization levels must be introduced.
Basic four levels of formalization proposed by the authors pertain to particular
forms of knowledge describing given product or process in the VR application. These
are the following:
1. Text form – descriptive knowledge (text messages, descriptions etc.)
2. Multimedia form – raster graphics (textures, schemes etc.) and audio/film
3. Text form – knowledge of (a) structure (list of parts – BOM, object positions in 3D
space, colors) and (b) operation (options, values and their constraints, process
sequence with progress conditions, animations etc.)
4. 3D model form – geometry of product, tooling, environment and characters, along
with animations.
Usually level 1 and 2 are easy to achieve by simple means of reading appropriate
files from external folders rather than internal resources hard-coded in the application.
Level 3 is more difficult, as it requires introduction of dynamic hierarchy, positioning
and flexible programmed logic. Level 4 is the most difficult to perform properly, as 3D
models in a traditional work scheme are usually known and prepared well before
programming other functionalities and certain operations (e.g. texture mapping,
lightmap baking) can have limitations if performing them during the runtime.
The knowledge acquired for a given application can be described quantitatively and
qualitatively. In terms of quantity, a typical unit of knowledge is composed of a text, a
2D graphics, a 3D model (optionally with animation) and optionally an audio. This can
constitute, e.g., a single part or subassembly of a designed machine or a single block in
a training program. Qualitatively, description of achieved levels should be marked, as
e.g. 1, 2 and 4 meaning that the acquired knowledge contains simple descriptions (text),
graphics and 3D models but does not contain metadata on structure and/or operation.
2.3 Knowledge Management Tools

Formalized knowledge management by manual file edition is a labor consuming task,
which also required advanced knowledge of application mechanisms and it can also
generate malfunctions (e.g. by a simple typo in a name of object inside a metadata file).
That is why for effective knowledge management there is a need of building an
additional application (module of a system). Its main functionality can be compared to
CMS (Content Management Service) applications used for edition of WWW pages.
Aims of its work are as following:
– acceleration of process of knowledge edition when reconfiguring or adding new
contents to a VR system – by automation of edition of files stored in the knowledge
base,
– elimination or serious limiting of risk of making an error while editing formalized
knowledge in the application,
– facilitation of management of data exported by the visualization module as a result
of performed activities by a user (e.g. product configuration report, virtual training
statistics),
– in case of dispersed (network) system architecture, i.e. many remote instances of the
visualization module – propagation of the updated knowledge base to all the VR
system instances, with use of cloud mechanisms.
Application for knowledge management in an open VR system may have one of the
two following forms:
– windowed 2D application without the 3D mode,
– 3D application with a graphical user interface.
The former is good for simple metadata and data management, when there is no
need to preview 3D data from inside the management application. The latter is required
when advanced 3D scenes are composed in the VR system and lack of 3D preview
would not allow their effective placement. The latter type of knowledge management
software is well known in the video game industry as the so-called “level editors”.
3 Selected Use Cases and Results

3.1 Presenter of Upholstered Furniture
The Presenter of Upholstered Furniture (PMT) is an application for virtual design and
ordering of single pieces of upholstered furniture. The application is of “projective VR”
class, with a given product presented on a large screen with touch-based controls. It is
meant for use by a client in a retail shop. The application has two main client-side
modules: module for selection of product and module for presentation of the selected
product (Fig. 3). These modules were created in the EON Studio software.
Fig. 3. The presenter of upholstered furniture (PMT) – presentation view
Regards knowledge formalization in the PMT application, it was assumed at levels

1, 2 and 3b, with omitting the levels 3a and 4. This was due to the manufacturer not
having 3D models of their products, so all the model preparation was outsourced to the
authors of the application, thus allowing hard-coding.
The following knowledge about the product was described in a formalized way:
– names and types of furniture,
– icons of products,
– photographs and movies,
– prices,
– textures,
– colors,
– options and values (amount of options and amount of textures/colors for a given
option for a given piece of furniture, with upper limit),
– contents of product specification for ordering (names of colors, contents of message
sent to user etc.).
The PMT application folder structure with description is presented in Fig. 4. Fig-
ure 5 presents a knowledge description model for automated generation of list of pieces
of furniture in the first module of the application. On the basis of code name of a
furniture, being direct name of its icon, the algorithm loads appropriate 2D and text
data, as well as creates appropriate link after clicking on the furniture, to launch the
second module (visualization of a single piece). The other metadata allow sorting and
searching for a given furniture or group of it.
Figure 6 presents examples of metadata describing changeability of textures and
colors for a given surface of a given piece of furniture. The file contains a list of
references to texture files or particular colors and their icons. On the basis of presence
of appropriate number of such defined files and their contents, the visualization module
generates an appropriate number of dropdown lists in GUI, allowing configuration of a
product.
Fig. 4. PMT application folder structure
Fig. 5. PMT application – formalized list of objects (part of metadata file)
Similarly, other interactive elements are defined. A number of metadata text files
exist for a given furniture piece, on that basis all the knowledge is presented (names,
dimensions, prices etc.).
Fig. 6. PMT application – formal description of color changeability (left) and textures (right) for
different surfaces of a single object (part of a metadata file)
Structure of the metadata files is strictly formalized, with specific separators used
(commas, semicolons, new line markers etc.) along with appropriately shortened folder
references. User of the system can decide what products are visible in the application,
how are they classified, what can be changed in them, what is their price and what other
important information (stored in form of graphics and movies) is attached to them. New
textures and colors, as well as multimedia, can be added to the system anytime.
Because the formalization is not complete (no level 4), adding entirely new products is
only possible from the development level in the software, what was intended. However,
the procedure of creating a new virtual product on the basis of its 3D model is simple
and it does not require any coding whatsoever, but rather a set of simple operations,
such as model import and creating references in appropriate places, as well as changing
some metadata inside a visualization file.
Additional 2D application was created using the Visual Basic language, for easy
management of knowledge inside the application and its propagation into many
instances of the system present in retail stores. The so-called administrative application
(PMT Administrator) is presented in Fig. 7. It allows changing virtually all the
Fig. 7. Knowledge management tool in the PMT application

formalized knowledge, but in a user-friendly manner, not allowing to introduce any

mistakes that would generate an error.
Studies on the PMT application allow to state, that it allows generating few hundred
to few billion of different variants of a given product (a single variant means a unique
specification). Single variant takes less than a minute to prepare. Therefore, a new
quality was obtained, which is capability of rapid generation of visualization and
specification for ordering, what makes a process of building new individualized variant
of a product easier on the manufacturer’s part – there is no need of manual creating of
variant documentation by a salesman. The application was implemented in retail stores
throughout Poland, with an initial package of 39 products and more than 200 different
textures and colors. The development process took approximately one year, with a
team of five people. Over the course of application run in the stores, capability of
efficient adding of contents was proven, as the application was later expanded by 29
new products and approximately 40 textures, in a time of three months, realized by two
people: one of original developers and one untrained student, both using the admin-
istrative application. Further development work also focused on creating and imple-
mentation of a prototype system of design of wooden furniture, intended for internal
use in the company [18].
3.2 Virtual Skill Teacher

The Virtual Skill Teacher (VST) is a VR platform for learning practical, technical
skills, usually related with operating working vehicles (such as the forklifts, which are
the main case), machines, realizing tasks in a certain production process, etc. The key
ideas are as following [19]:
– the course is the main entity, it is divided into lessons and exercises, which are
further divided into blocks,
– lessons are theoretical and contain a number of graphical, text, audio and interactive
3D info presented in immersive space,
– exercises are sets of practical tasks – they require certain actions in a specified time
(such as: drive from place A to B or bring object from one place to another and use
it properly) in order to be passed,
– lessons and exercises are divided into blocks, containing certain, distinguished
forms of knowledge (described below).
– each course has a set of questions, the exam mode is in-built and also performed in
the immersive space.
The contents of lessons have the following form:
– text – descriptions of problems, especially regarding information difficult to present
in a visual form (names of standards, formulas etc.), as well as instructions,
– audio – complementary to the text content, read by a lector and launched by
interaction with a character of virtual teacher, who is always present in the virtual
space,
– 2D graphics – pictures, schemes, diagrams, infographics illustrating blocks,
– 3D content – interactive models of objects (machines and devices, elements of work

environment etc.), along with animations, illustrating a current block.
The most important constant elements of the lesson scene are:
– virtual teacher – an animated character helping the user, replacing a living teacher,
– flipchart – space in the center of the scene, where 2D content is displayed,
– environment – e.g. a hall – 3D background space, that the user can navigate.
The user can move freely using three means: on own legs (only short distances, due
to HMD cable length and workspace limitation), by use of joystick and by use of
teleporting points, located in the environment. Interaction is done by gaze & click
mechanism or by using a gaming driving wheel to move the vehicle in the practical
exercises. Figure 8 presents a screen from the finished application, showing the most
important elements of a single lesson.
Fig. 8. Basic elements of a lesson in the VST system
The application, made in Unity engine, is thought of as a platform, meaning that it

can be complemented with other courses. The current version contains only one course
(forklift) but other courses are planned (welding, gantry crane operation etc.). The
application studies allowed to state that it is very effective, allowing 90% of shortening
of learning time compared to the original, traditionally performed course [19].
The VST system allows flexibility in content management – it allows easy edition
and adding new knowledge, i.e. new courses, lessons etc. with minimal or no pro-
gramming skills necessary. The access to most knowledge is open and metadata are
stored in form of text files of open structure. Exemplary structure of an exercise block
is presented in Fig. 9. A separate Windows application (VST Administrator) was built
for knowledge management. The formalization level is similar to the previous case –
levels 1, 2 and 3b were assumed, with 3a and 4 levels omitted. The level 4 was
unnecessary in the first version of the application, but it will be introduced in further
installments (the application is still in the development phase – 3 prototypes were

created and tested so far).
Fig. 9. Structure of formalized knowledge in the VST system [19]
The studies on effectiveness of knowledge management in the application consisted

in two phases:
1. Building prototype 1 containing approx. 40% of contents of the application, without
special knowledge management tools.
2. Building prototype 2 containing full contents of the application, with use of dedi-
cated knowledge management tools.
In total, the forklift operator course consists of 9 theoretical lessons and 4 practical
exercises. This makes for total of 154 knowledge blocks. Prototype 1 contained 3
lessons and 1 exercise, which makes for total 56 knowledge blocks. The team assigned
to building both prototypes was the same and consisted of 2 programmers and one 3D
graphics specialists, with consultations with two experts (instructors). The results
regarding time of building the system by blocks are presented in Table 1, while total
time consumption of building the system is presented in Table 2.
Table 1. Knowledge block adding – efficiency ratio comparison in the VST system.
Phase No. of blocks Work days Efficiency ratio (blocks per day)
Prototype 1 56 75 0.75
Prototype 2 98 25 3.92
Sum/mean 154 100 1.54
Table 2. Time consumption of building the VST system.

Stage Activity VR programmer 3D graphics specialist
work days work days
1 Acquisition of knowledge 5 20
2 Formalization – 1 10 3
3 Implementation (programming 60 2
& tests) – 1
4 Formalization – 2 15 3
5 Implementation (programming 5 2
& tests) – 2
6 Final improvements 10 -
Total: 105 30
The results comparison between prototype 1 and 2 clearly show, that introduction
of knowledge management procedures (namely, formalization of knowledge and
dedicated tools for simple knowledge edition) allowed to increase rate at which sub-
sequent blocks of knowledge are added to the application. Advantages of open access
to knowledge are clearly visible when comparing times of stages 2 & 3 with stages 4 &
5 (twice more content done in three times shorter total time). Apart from this, entirely
new quality was obtained, of open access to courses, lessons and their blocks, for future
edition or adding new courses outside the programming engine.
4 Summary
The new approach to storing and editing knowledge in VR applications created for the
production industry, presented in the paper, was proven to be effective. It requires
changing some of traditional paradigms of VR application development process, but in
exchange a capability of building open, manageable and easily extendable solutions is
reached. These solutions can be effectively implemented in production companies to
routinely support processes of design, production and employee training.
The proposed formalization approach requires considerably more programming in
the initial stages of development, so the presented ideas should be put into practice only
when there is a direct need for it and scale of a given system (number of products,
variants, complexity of processes etc.) justifies it – simple one-time visualizations
should be still made in a traditional way, with hard-coded knowledge. The key to
proper implementation of industrial VR is always performing of identification and
justification stages, that should answer the question of what type of application is
needed, who will use it and to what degree can it help.
The authors have already made and implemented a number of open VR systems in
various industries. The further research, carried out continuously, focuses on building
more complex solutions, with full formalization, and testing their prolonged use in
given industries, checking for efficiency and improving various aspects of the presented
methodology.
Acknowledgments. The research work reported here was partially financed by Polish Ministry
of Development (UE funds: Operational Program Knowledge Education Development) and
Polish Ministry of Science and Higher Education.
References
1. Fogliatto, F.S., da Silveira, G.J., Borenstein, D.: The mass customization decade: an updated
review of the literature. Int. J. Prod. Econ. 138(1), 14–25 (2012)
2. Tanner, S.: The use of virtual reality at Boeing’s Huntsville laboratories. In: Proceedings of
IEEE Virtual Reality Annual International Symposium (1993)
3. Hayden, S.: Volkswagen Group to Train 10,000 Employees in VR This Year, Road to
VR (2018). https://www.roadtovr.com/volkswagen-group-to-train-10000-employees-in-vr-
in-2018/
4. Brake, T.: Virtual Reality Collaboration at Volkswagen, Virtual Gurus (2017). http://www.
virtualgurus.com/insights-and-resources/virtual-reality-collaboration-volkswagen
5. Obermair, F.: MAN and the University of Applied Science Upper Austria Design a Truck
Assembly Line using ESI’s IC.IDO - ESI Success Stories, pp. 1–2 (2015)
6. de Sa, A.G., Zachmann, G.: Virtual reality as a tool for verification of assembly and
maintenance processes. Comput. Graph. 23(3), 389–403 (2001)
7. Jimeno, A., Puerta, A.: State of the art of the virtual reality applied to design and
manufacturing processes. Int. J. Adv. Manuf. Technol. 33(9–10), 866–874 (2007). https://
doi.org/10.1007/s00170-006-0534-2
8. Martín-Gutiérrez, J., Mora, E., Añorbe-Díaz, B., González-Marrero, A.: Virtual technologies
trends in education. EURASIA J. Math. Sci. Technol. Educ. 13(2), 469–486 (2017)
9. Gulec, U., Yilmaz, M., Isler, V.: A literature survey: is it necessary to develop a new
software development methodology for virtual reality projects? J. Univ. Comput. Sci. 23(8),
725–754 (2017)
10. Cecil, J., Kanchanapiboon, A.: Virtual engineering approaches in product and process
design. Int. J. Adv. Manuf. Technol. 31(9–10), 846–856 (2007)
11. Choi, S.S., Jung, K., Noh, S.D.: Virtual reality applications in manufacturing industries: past
research, present findings, and future directions. Concurr. Eng.-Res. Appl. 23(1), 40–63
(2015)
12. Abulrub, A.H., Attridge, A., Williams, M.A.: Virtual reality in engineering education: the
future of creative learning. Int. J. Emerg. Technol. Learn. 6(4), 4–11 (2011)
13. Wang, X., Ong, S.K., Nee, A.Y.C.: A comprehensive survey of augmented reality assembly
research. Adv. Manuf. 4(1), 1–22 (2016)
14. Palmarini, R., Erkoyuncu, J.A., Roy, R., Torabmostaedi, H.: A systematic review of
augmented reality applications in maintenance. Robot. Comput.-Integr. Manuf. 49, 215–228
(2018)
15. Calvo, I., Lopez, F., Zulueta, E., Gonzalez-Nalda, P.: Towards a methodology to build
virtual reality manufacturing systems based on free open software technologies. Int.
J. Interact. Des. Manuf. IJIDEM 11(3), 569–580 (2017)
16. Gorecky, D., Khamis, M., Mura, K.: Introduction and establishment of virtual training in the
factory of the future. Int. J. Comput. Integr. Manuf. 30(1), 182–190 (2017)
17. Górski, F.: Building virtual reality applications for engineering with knowledge-based
approach. Manag. Prod. Eng. Rev. 8(4), 64–73 (2017)
18. Zawadzki, P., Górski, F., Buń, P., Hamrol, A., Kuczko, W., Wichniarek, R.: Concept design
of furniture with use of virtual reality applications and knowledge based engineering (in
Polish). Innovations in Management and Production Engineering 2016. Publishing House of
the Polish Society for Production Management (2016)
19. Górski, F., Zawadzki, P., Buń, P.: Virtual skill teacher – platform for effective learning of
technical skills. In: Proceedings of the International Conference of the Virtual and
Augmented Reality in Education (2018)
The Automated Acquisition of Expert
Knowledge Using a Service Department
as an Example
Justyna Patalas-Maliszewska1(&), Adam Dudek2,

and Sławomir Kłos1
1
Institute of Computer Science and Production Management,
University of Zielona Góra, Zielona Góra, Poland
j.patalas@iizp.uz.zgora.pl
2
Institute of Technical Science, University of Applied Science in Nysa,
Nysa, Poland
Abstract. In this article, an example of the use of ontology for the automated
acquisition of expert knowledge in a service department is presented. The first
part of the article presents an overview of the literature on knowledge acqui-
sition methods and an evaluation of knowledge tools. The model for the
Automated Acquisition of Expert Knowledge (AA-EK) is then presented and a
formal representation of it, using ontology, is provided. The last part illustrates
how the AA-EK model is used, using a real-time case study.
Keywords: Expert knowledge Knowledge acquisition

Expert knowledge formalisation
1 Introduction
Knowledge can be divided into explicit knowledge and tacit knowledge. In addition,
the transformation of one type of knowledge into another type of knowledge can be the
first step in the creation of new knowledge within a company [1]. Tacit knowledge is
inter-connected with persons, in the form of know-how and is used in order to achieve
work goals [2–4]. Tacit knowledge can be treated as the expert knowledge within a
company. This can be perceived as a competitive advantage within an organisation
because it cannot be replicated by competitors.
The aspect of the representation of expert knowledge should be the main func-
tionality of a knowledge-based system, as well a critical step in the development of any
system [5, 6]. But the first step in the codifying of expert knowledge must be the
knowledge acquisition. Where the planning of the acquisition of knowledge is for a
system, the forms of its representation, the methods for its recording and its indexation,
according to standardised rules, should be ascertained thoroughly, so that the knowl-
edge is useful in ongoing processing. Therefore in order for the expert knowledge
acquired to be useful to an enterprise requires its formalisation. The ontology was
proposed for the formalisation of acquired expert knowledge.

https://doi.org/10.1007/978-3-030-18715-6_10
120 J. Patalas-Maliszewska et al.
So, in this paper, we develop the model for the Automated Acquisition of Expert
Knowledge, (AA-EK), based on the example on a manufacturing company and a
service department. We attempt to codify acquired expert knowledge, based on the
example of the knowledge of service workers in a Polish manufacturing company.
Expert knowledge is acquired from workers with the help of devices such as smart-
phones. The spoken recordings obtained are then converted into the written form using
Automatic Speech Recognition technology (ASR). Speech signals exhibit many vari-
ations, two different people saying the same sentence invariably produces different
results. Moreover, the use of the Speech Recognition technology, on their own, is
insufficient for recognising acquired expert knowledge. We propose ontology for tasks
completed by service workers, extending it with the Finalised Word Dictionary (SSW),
in order to transform the workers’ text into a conceptualised text. Finally, we evaluate
the Automated Acquisition of Expert Knowledge model (AA-EK) with regard both to
the sentence (step) level task and the document (service procedure) level task. The
results of the experiment illustrate the effectiveness of our work.
2 Related Work
Enterprises want to invest in technologies, inventions, products, and ideas [7]. Expert
knowledge more and more often determines the company’s potential. In selecting the
expert knowledge acquisition method - and also the tacit knowledge acquisition
method, it can be seen that knowledge can be acquired in a variety of forms; this
requires formalisation for further processing. Leu and Abbass conducted a thorough
review of the literature on knowledge acquisition methods [8]. They proposed a
division of methods from the point of view of the type of agency involved in the
knowledge acquisition exercise. The above cited authors chose a process-tracing
method, viz., the Human Agent Category method, based on this classification. This
method was also selected for the purposes of their research. Using this method, the
following techniques were identified: verbal reports, non-verbal reports, protocol
analysis, decision analysis and cognitive walkthrough [9]. The use of the knowledge
acquisition methods should attempt to change expert knowledge, into explicit
knowledge.
In our proposed model, namely AA-AK, we state that workers comment upon
completed tasks; this knowledge is then recorded on a smartphone with the recorded
audio then being automatically converted into text, using the ASR technique. In order
that acquired expert knowledge is to be of any future use to other employees, it is
necessary to formalise it. Ontological approaches are also used in the analysis of audio
materials; in our case, the formalisation of knowledge is derived from the service
department in a manufacturing company [10] or, again, as a document search system
which describes the ontological documentation indicated. Among the languages
The Automated Acquisition of Expert Knowledge Using a Service Department 121
currently used in formal ontology, the languages derived from a predicate calculus can
be indicated; these include, Ontolingua [11], Flogic [12, 13], using the logic of the first
order – OCML [14–16], based on descriptive logic – LOOM [17], or those languages
derived from semantic webs - SHOE [18, 19], RDF [20–22] or one of the most popular
languages, using XML - OWL [23].
The problem with using an ontological approach is the complication brought about
by the separation of knowledge- in the case of complex language flexion -and the lack
of high efficiency in programmes that translate text from recordings into the written
form. In the case of expert knowledge, complete automation of the knowledge veri-
fication process, such as the construction of universal tools, is not possible. In practice,
dedicated solutions for specific applications are used, based on the aforementioned
simulators, virtual reality or utility -serious- games. So, in our AA-AK model, we
established an ontology for service procedures, realised by service workers in the
manufacturing company, in the service department and extended it with the Finalised
Word Dictionary (SSW).
3 The AA-EK Model
In our AA-EK model (Fig. 1), we state that knowledge from workers within a man-
ufacturing company (expert knowledge so-called) is acquired with help of devices such
as smartphones. The recordings obtained are converted from the verbal form to the
written form, using Automatic Speech Recognition technology (ASR). For the formal
representation of knowledge, an ontology was developed for selected service proce-
dures in a manufacturing company. In the AA-EK model, the ontology was created in
co-operation with a specialist technician from the service department.
The written record of the execution of the service procedure, expressed in the form
of a text document is expressed in naturally spoken Polish. Polish is characterised by
ambiguity and extensive flexography, which makes processing and understanding it
quite difficult to automate. It is also impossible to effectively evaluate received
knowledge, in such a form, using only basic NLP (Natural Language Processing) and
Shallow Text Analysis, so-called, based on statistics and recognition of the main parts
of sentences [24, 25]. This is due to the fact that both entities and rules can occur in
many different forms, such as times, modes, cases, etc., while maintaining the same
meaning. As has been demonstrated for the set of documents selected, the most
common expressions were mostly auxiliary words which, in Polish, play the rôle of
prepositions, conjunctions, pronouns and adjectives but which are not proper parts of
the sentence, as such.
Stage 1: Create the ontology for a chosen service procedure within

a manufacturing company
Stage 2: Extend the ontology with the Finalised Word Dictionary

(SSW)
Stage 3: Acquire the expert knowledge: verbal recording by service

workers of completed activities (service procedures)
Stage 4: Convert the spoken recordings into textual records, using

the Google Speech Api technique
Stage 5: Determine the steps most similar, in the text (procedure),

as defined in the ontology
Stage 6: Determine the identifier of the procedure, as defined in the

ontology
Stage 7: Compare the expert knowledge acquired with the currently

extant one
Fig. 1. The AA-EK model.
In the AA-EK model, the ontology describes the correct course of service proce-
dures and represents them by sequences of steps. Each step can be realised in different
variants and each of them is described by the characteristics of the steps, i.e. the names
of actions, such as, basic actions and generalised actions; the names of sub-assemblies;
the names of tools; the names of connecting elements and the names of consumables.
This ontology was extended by the Finalised Word Dictionary (SSW), so-called, which
contains important concepts concerning the domain of tasks for which the acquisition
of knowledge was conducted. Finally, the SSW glossary includes about 200 basic
forms, some 400 synonyms and around 11,000 flexible forms.
The knowledge, acquired from workers in the written form, using Automatic
Speech Recognition technology (ASR) as natural language, must be transformed into a
simplified form so that it can be formalised further; it is proposed to implement this
process on the basis of an algorithm:
1. Remove from the records, under analysis, any words from the Stop List, so-called,
such as, prepositions, pronouns, conjunctions and particles.
2. Replace all instances of flexible forms from the SSW dictionary with their basic
forms.
3. Specify the main form, such as the names of actions, sub-assemblies, tools, con-
necting elements and consumables, for each of the basic forms of the synonym, as
defined in step 2.
4. Determine the affiliation to the corresponding group of generalised functions in the
ontology, for each of the actions in step 3.
5. Describe the positions of all generalised actions, identified in step 5, in the log
analysed.
In order to be able to compare the knowledge acquired with that currently extant, it
must be properly formalised, that is, expressed by a common, conceptual system -
ontology. Building - and then using - ontology allows unambiguous, terminological
terms to be used and facts from individual fields of knowledge to be recorded, using a
fixed conceptual model and formally specified semantics that can be processed by
machines. Using the proposed ontology, an assessment of acquired knowledge can be
made by comparing it with knowledge previously held and regarded as master
knowledge. In the next chapter, a case study for using the AA-EK model in the service
department of a manufacturing company, is presented.
4 The Use of the AA-EK Model in a Manufacturing

Company – A Case Study
In this chapter, an example of using the AA-EK model, in a service department, is

presented.
Using the SSW glossary with inflectional forms, it is possible to transform the
recordings, expressed in naturally spoken Polish, into the form expressed by concepts
defined in ontology. It should be emphasised that the elements of speech that do not
have their forms in the SSW glossary are not included in further analysis. This is
illustrated by the following example.
… przed rozpoczęciem demontażu warto przeczyścić szczotką drucianą zacisk oraz
jarzmo usuwamy sprężynę zabezpieczającą klocek hamulcowy przy użyciu kombinerek
usunąłem zaślepki sworzni zacisku śrubokrętem i odkręcamy te sworznie grzechotką z
imbusem …
The written record, compiled from the voice recordings, using the SSW glossary, is
analysed using the following rules:
• Where, in the Polish textual records, there are prepositions, pronouns, conjunctions
and particles, then delete.
• Where there are flexible forms in the textual records, then replace them with their
basic forms from the SSW dictionary.
• For each of the basic forms, specify the main form.
• Determine the affiliation to the ontology.
• For each “names of actions” define the sub-sets and include the sub-assemblies and
tools connecting the elements and consumables.
• For each sub-set:
– If, in the sub-set, there is a minimum of 1 sub-assembly name, or 1 tool name, or
1 connecting element name, or 1 consumables name, then add this sub-set to the
result.
– If, in the sub-set, there is not a single, sub-assembly name, as a minimum, or a
minimum of 1 tool name, or a minimum of 1 connecting element name, or a
minimum of 1 consumables name, then pass over this sub-set.
As a result of the implementation of the rules, a sequence of step characteristics (i.e.
the names of actions, viz., basic actions and generalised actions; the names of sub-
assemblies; the names of tools; the names of connecting elements and the names of
consumables) is determined, each of which is expressed in a form consistent with the
ontology.
Based on the example above, we can illustrate a sequence of step characteristics:
• Characteristic1 = [cleaning (czyścić), cleaning (czyszczenie), brake caliper (zacisk
hamulcowy) brake caliper yoke (jarzmo zacisku), none, hand wire brush (szczotka
druciana ręczna), none].
• Characteristic2 = [remove (usunąć), removing (usuwanie), brake block spring
(sprężyna zabezpieczająca klocek hamulcowy), none, pliers (kombinerki uniwer-
salne), none].
• Characteristic3 = [remove (usunąć), remove (usuwanie), clamp (zacisk), blanking
(zaślepka) pin (sworzeń), flat screwdriver śrubokręt płaski, none].
• Characteristic4 = [unscrewing (wykręcać), unscrewing (wykręcanie), none, pin
(sworzeń), rattle standard (grzechotka standardowa) allen wrench (klucz imbu-
sowy), none].
Based on the ontology created, using SSW, the formal representation of expert,
acquired knowledge can be established for the defined step characteristics:
• Characteristic1 = [1, 1, 32 33, none, 48, none].
• Characteristic2 = [22, 15, 27 30, none, 50, none].
• Characteristic3 = [22, 15, 32, 19 20, 53, none].
• Characteristic4 = [22, 15, none, 20, 29 53].
As the result of the comparison of expert knowledge, we are able to obtain a
benchmark for verifying the accuracy of the performance of the service tasks. In order
to be able to compare the knowledge acquired with the reference model, a common
form was proposed so that it would be possible to determine the correctness of the
service procedure. In the case of irregularities, it is also possible to consider whether
the method for realising it is actually incorrect, or whether it is a new, previously
unknown -albeit valid- variant of its realisation.
The use of this AA-EK in the manufacturing enterprise will allow the service
manager to gain knowledge regarding the degree of similarity with a procedure - or
procedures - which were previously defined as having been correctly completed, when
he/she knows which of the procedures was performed.
On this basis, the manager makes the final decision as to whether the analysis of the
task actually completed is correct.
5 Conclusions
In reviewing the literature on expert workers, it is evident that, as intangible sources

within a company, they play a crucial rôle in a company’s development processes.
Unfortunately, however, their knowledge is very difficult to absorb.
We have described the AA-EK model for the acquisition of expert knowledge,
based on the example on knowledge of service workers in a manufacturing company.
Our framework is applied both to the steps in the procedure level task and to the
document (procedure) level task. For the step level task, we test our model by a step
with ID 62 (screw-in screw of the yoke clamp/wkręcanie śruby jarzma zacisku) in
different variants. For the procedure level task, our model will be tested on procedure
classification.
In our further work we are planning to apply the artificial neural networks for the
comparison of the record of the service procedure to the reference procedures. The
model will consist in classifying the characteristics of the maintenance procedure steps
that were extracted from the recording of the procedure by the service worker. These
characteristics will be classified according to the degree of their similarity to the
characteristics of the model steps defined in the ontology. As a result of the network
operation, a sequence of steps will be obtained that are the most similar to those
described in the ontology.
Also, our approach to the acquisition of expert knowledge, presented here, can be
employed for verifying the skills levels of specialists and possibly, even for verifying
the skills of those who are learning to do these tasks.
References
1. Nonaka, I., Takeuchi, H.: Creating Knowledge in the Organisation. Poltext, Warsaw (2000)
2. Polanyi, M.: The Tacit Dimension. Anchor Books, Garden City (1967)
3. Noh, J.B., Lee, K.C., Kim, J.K., Lee, J.K., Kim, S.H.: A case-based reasoning approach to
cognitive map-driven tacit knowledge management. Expert Syst. Appl. 19(4), 249–259
(2000)
4. Brockmann, E.N., Anthony, W.P.: Tacit knowledge and strategic decision making. Group
Organ. Manag. 27(4), 436–455 (2002)
5. Hoffman, R.R., Shadbolt, N.R.A., Mike, B., Klein, G.: Eliciting knowledge from experts: a
methodological analysis. Organ. Behav. Hum. Decis. Process. 62(2), 129–158 (1995)
6. Tang, F., Jifeng, M., MacLachlan, D.L.: Implication of network size and structure on
organizations knowledge transfer. Expert Syst. Appl. 34(2), 1109–1114 (2008)
7. Araújo, A.F., Varela, M.L.R., Gomes, M.S., Barreto, R.C.C., Trojanowska, J.: Development
of an intelligent and automated system for lean industrial production adding maximum
productivity and efficiency in the production process. In: Hamrol, A., Ciszak, O., Legutko,
S., Jurczyk, M. (eds.) Advances in Manufacturing. Lecture Notes in Mechanical
Engineering, pp. 131–140 (2018)
8. Leu, G., Abbass, H.: A multi-disciplinary review of knowledge acquisition methods: from
human to autonomous eliciting agents. Knowl.-Based Syst. 105, 1–22 (2016)
9. Cooke, N.J.: Varieties of knowledge elicitation techniques. Int. J. Hum.-Comput. Stud.
41(6), 80–849 (1994)
10. Patalas-Maliszewska, J., Dudek, A.: A model of tacit knowledge transformation for the
service department in a manufacturing company: a case study. Found. Manag. Int. J. 8(1),
175–188 (2016)
11. Gruber, R.T.: Ontolingua: A Mechanism to Support Portable Ontologies. Stanford
University (1992)
12. Neumayr, B., Schuetz, C., Jeusfeld, M.A., Schrefl, M.: Dual deep modeling: multi-level
modeling with dual potencies and its formalization in F-Logic. Softw. Syst. Model. 17(1), 1–
36 (2016)
13. Slimani, T.: Ontology development: a comparing study on tools, languages and formalisms.
Indian J. Sci. Technol. 8(24), 1–12 (2015)
14. Negri, E., Fumagalli, L., Garetti, M., Tanca, L.: A review of semantic languages for the
conceptual modelling of the manufacturing domain. In: Proceedings of the XIX Summer
School “Francesco Turco” Industrial Mechanical Plants-Senigallia, Italy (2014)
15. Motta, E.: An overview of the OCML modelling language. In: The 8th Workshop on
Methods and Languages (1998)
16. Xiaomeng, S., Ilebrekke, L.: A comparative study of ontology languages and tools. In:
International Conference on Advanced Information Systems Engineering, pp. 761–765.
Springer (2002)
17. Gomez-Pérez, A.: Ontological engineering: a state of the art. In: Expert Update: Knowledge
Based Systems and Applied Artificial Intelligence, pp. 33–43 (1999)
18. Heflin, J., Hendler, J., Luke, S.: SHOE: a knowledge representation language for internet
applications (1999)
19. Horrocks, I., Patel-Schneider, P.F., Van Harmelen, F.: From SHIQ and RDF to OWL: the
making of a web ontology language. Web Semant.: Sci. Serv. Agents World Wide Web 1(1),
7–26 (2003)
20. Czarnecki, A.: IT technologies used in the design and implementation of ontologies. In:
Orłowski, C. (ed.) Management of Information Technology: State and Prospects of
Development, pp. 83–94. Pomeranian Scientific and Technical Publisher (2006). (in Polish)
21. Czajkowski, K., Trela, T.: Semantic Web–standard, tools, implementations. Studia
Informatica 33(2A), 379–393 (2012)
22. Mach-Król, M.: Knowledge and ontologies in temporal intelligent systems. Inf. Econ. 1(35),
28–35 (2015). (in Polish)
23. Goczyła, K.: Ontologies in information systems, EXIT (2011). (in Polish)
24. Kempa, A.: Application of an extended inference methodology based on case-textual CBR
in working with text documents, Organizational Support Systems (2005). (in Polish)
25. Hernes, M., Maleszka, M., Nguyen, N.T., Bytniewski, A.: The automatic summarization of
text documents in the cognitive integrated management information system. In: Federated
Conference on Computer Science and Information Systems (FedCSIS), pp. 1387–1396.
IEEE (2015)
Virtual Simulation of Machine Tools
Damian Grajewski1(&), Filip Górski1, and Zoran Pandilov2

1
damian.grajewski@put.poznan.pl
2
University “Sv. Kiril i Metodij”-Skopje, Skopje, Republic of Macedonia
Abstract. The article presents the use of VR tools for creating virtual simu-
lations in which construction and functionality of production machines can be
observed. As examples, the CNC machine feed drive and the Delta robot were
presented. A virtual CNC machine was used to visualize its operation in the
EON Studio environment. This visualization enabled a general analysis of the
machine’s operation and construction, visualization with various tools mounted
in the spindle, as well as execution of the tool movement according to the loaded
simple G-code. In turn, virtual model of Delta robot, in addition to the visual-
ization of the action, was used to simulate the shape of virtual objects.
Keywords: Virtual Reality Virtual simulation Visualization Machine tools
1 Introduction
Modern Virtual Reality (VR) techniques find their application in many phases of the
product life cycle. VR solutions available on the market allow for the preparation of
immersive, interactive environments with a high degree of realism. Virtual models are
analyzed throughout the entire product life cycle, starting from the concept phase in
which the future product can be visualized - then the so-called virtual prototype, and
ending with the phase of withdrawal from life - creates, among others, interactive
recycling instructions [1].
Virtual models are placed in created virtual scene in order to [2]:
– testing, improvement and development of the future product in the design phase
(e.g. improvement of construction, ergonomics, simulation of operation),
– optimization of the remaining phases of the life cycle (including production
preparation - virtual production processes, production - virtual assembly
instructions).
When it comes to supporting virtual techniques in the design phase, the first results
of research that show the measurable benefits of using VR in industrial design,
appeared in the late 80’s. Researchers [3] from the University of North Carolina have
prepared a solution called Walkthrough, which was used to create virtual simulations of
buildings (visualization of building forms, roof, façade and room layout). The digital
model of the building was prepared on the basis of documentation provided by the user

https://doi.org/10.1007/978-3-030-18715-6_11
128 D. Grajewski et al.
(customer). The analysis of the interactive version of the designed building allowed for
the detection and improvement of errors before the actual construction works.
One of the main advantages of using VR solutions in design is the potential to
reduce project costs [4–6]. Entities participating in the product life cycle (including
designers, constructors, technologists, marketing experts) have the opportunity to
familiarize themselves with the virtual prototype of the future product and the envi-
ronment in which it will be used. Thanks to this, it is possible to detect potential errors
(e.g. in construction) in the early phase of the project, while in traditional design you
had to build a physical prototype to subject it to the necessary analyses.
The use of VR tools and techniques in the design and prototyping of products was
the subject of works [7–10] implemented in the Virtual Design Laboratory, whose
members are the authors of this publication.
The paper presents the use of VR tools for creating virtual simulations in which
construction and functionality of production machines can be observed.
2 Materials and Methods
2.1 Virtual Reality in Design

The main task of CAx (Computer Aided, where “x” stands for Design, Manufacturing,
Engineering etc.) systems is to support the shaping of product geometry, its visual and
technical features. CAx systems do not deal with the interaction of the modeled product
with other objects, in addition to very basic operations, e.g. collision testing in
assembly modules. Do not confuse the digital design of new products or processes in
CAx systems with virtual design and prototyping - the difference is primarily
immersion or semi-interactive interaction [11].
Virtual Reality techniques, and in particular - virtual prototyping used in this
context, therefore broaden significantly the spectrum of application of the model cre-
ated in the CAD (Computer Aided Design) environment, allowing the placement of a
digital model of the product in the presence of other virtual objects (in the virtual
scene). The most important feature of the virtual environment is the mapping of the
interaction of objects and their behaviors in response to user-triggered events, with the
user’s free interaction and immersion. Available 3D engines allow creating solutions so
realistic that it is possible to test and improve the product only in the virtual space,
without the need to create a physical prototype [1, 7]. The most important measurable
reasons for using the virtual prototyping technique in design support are primarily the
significant reduction of the functional evaluation costs of the new product [11],
shortening development time by as much as half [12] or potential increase of its quality
[13] and even environmental issues [14]. It is also possible to make quick decisions
about the future shape of the product and eliminate its weakness before it has a physical
form [15].
A product model existing only in a virtual environment, used throughout the design
process to perform tests and trials aimed at its improvement and development is called
a virtual model. It is a digital representation of a real product that can be used
throughout the product’s life cycle, in its appropriate places. The virtual model of the
Virtual Simulation of Machine Tools 129
product is used in many fields of industry, including Automotive, aerospace, nuclear

etc. industries. In these industrial branches, all stages of designing, testing and
improving the product as well as the preparation of its production are currently based
only on the virtual model, the product obtained in the virtual design process is com-
pletely prepared for production [5].
In the case of design support, a virtual product should be identified with a virtual
prototype, as long as iterative optimization of the construction takes place. Virtual
prototype, created on the basis of 3D CAD data or coming from 3D visualization
systems, are tested in a virtual environment and based on them makes structural
improvements until it is verified positively. Then the design phase can be considered
finished and go to further phases - preparation of production, production of a physical
prototype, etc.
2.2 Method of Creating a Virtual Reality Application

The most important component of the Virtual Reality system is the VR application,
which should be understood as a closed (in the programming, logistic and functional
sense) whole, functioning as an executable program (requiring a selected operating
system and using specific VR equipment). The main task of the creators of VR
applications is to create an interactive, three-dimensional environment that will provide
the user with the immersion and realism of simulation, as well as the interaction with
digital elements [16].
The two basic types of VR applications are applications oriented to a single object
(usually the virtual model of a specific product) and the environment (virtual scene).
The division of application types is shown in Fig. 1 [2]. From the point of view of the
topic of the article, the most important types of VR application applications include
virtual simulations of machines and devices [17].
Preparing an interactive VR application is a complex and time-consuming process.
The basic stages of creating a virtual application are shown in the diagram (Fig. 2),
taken from works [2].
The first stage is preceded by works related to [2]:
– selection of VR equipment (depending on the needs, e.g. providing the effect of
force feedback or full immersion),
– VR software selection (criteria: license cost, availability, ease of programming and
operation, compatibility with selected VR equipment),
– preparation of data for the application (3D models: products, items, infrastructure
elements, 2D data: textures, user interface graphics, predefined lighting and shading
maps).
The first step in the stage of visualization preparation is the transfer of digital data
(conversion to a form recognizable by a dedicated environment), most often using
external software that ensures the integrity of individual parts of 3D CAD models. The
imported models are assigned:
– appropriate textures (2D graphic patterns, placed on a 3D object to imitate the
appearance of the material, invoice, etc.),
Fig. 1. Examples of VR applications types [2].
Fig. 2. Stages of the VR application building process [2].
– normal maps and bounce maps (assigning individual pixels of a 3D image to the
vertices of a 3D mesh, in order to obtain a more realistic appearance of objects),
– so-called. material properties (a set of features describing the appearance of a given
object in a VR environment).
The scale and location of digital data in the virtual scene are also determined.
Subsequent works rely on appropriate lighting of objects and adding shading effects
(among others based on the location of objects and lights casting shadow). The last step
in the stage of visualization preparation is to develop a way of navigating the virtual
scene (camera position change control, e.g. a camera directed at the central point or the
so-called first person camera - user’s free mode).
The second, key stage in creating a VR application is programming interaction, i.e.
behavior of virtual objects. Most often these are specific movements, which are a
response to an event generated by the user (e.g. by clicking the mouse, pressing a
button on the keyboard, entering a defined collision area) or by another object [1].
Other examples of object behaviors are:
– dynamic change of geometry or change of appearance (e.g. different color, texture,
scale, cross-sections),
– deformation of 3D objects (so-called animations with model deformation),
– collisions (e.g. by adding collision objects).
Programming of specific behaviors is carried out using traditional programming
methods (C++, Visual Basic, Python) or visual (with numerous logical conditions to
handle events), consisting in defining connections between the so-called nodes
exchanging data of various types (Fig. 3).
Fig. 3. Visual programming (on the example of Unreal Engine) [16].
The third stage of building a VR application is work related to the construction of a

graphical user interface - GUI, which will work with selected interaction devices.
We’re talking about solutions based on:
– peripheral devices (e.g. mouse and keyboard),
– advanced electronic equipment (e.g. tracking systems, gesture recognition systems).
The main goal in the process of creating the user interface is always to achieve the
desired degree of intuitiveness and transparency for the purpose of running defined
functions in a virtual environment [2]. The last step of the third stage of building the
VR application is the testing phase in which you verify:
– the geometry of 3D models used in the simulation,
– correct operation of the GUI,

– functionality of the application as an integral whole,
– cooperation with specific VR equipment.
3 Examples of Virtual Simulations
3.1 CNC Machine

The typical CNC (Computer Numerical Control) milling machine feed drive consists of
many elements that influence on the machine important parameters like movement
stability and positioning accuracy. The design of complex CNC milling machine feed
drives requires laborious and time-consuming calculations supported by experimental
tests. Therefore the modelling, simulation and visualization techniques may be very
useful and may facilitate the design process. The simulation and visualization in VR
environment may show the week points of the drive. Moreover, the changing and
optimization of several drive control parameters may be tested and validated [10].
Complicated simulations were performed in Matlab-Simulink environment and
based on the obtained results, the drive functioning and influences of different drive
parameters were later visualized in VR system. What was more important, the visu-
alization could be based on the model simulations results and on that way we could
avoid complicated calculations of the real behaviour of the VR system. The visual-
ization of the whole machine tool using the CAD model of machine tool and models of
the feed drives developed in Matlab-Simulink and Sim Scape in EON Studio software
is presented on Fig. 4. Virtual model was used to simulate the real work of the whole
CNC machine tool (contouring operations, changing tools, etc.) This whole research
could be very useful to machine tool producers, because they could change their
practice of making very expensive real prototypes of the CNC machine tools, with
creating virtual prototypes of machines [10].
Fig. 4. Visualization of CNC portal machine tool in EON studio software [10].
3.2 Delta Robot

Just like in the case of a CNC machine, the Delta virtual robot model was used to
analyze the operation and construction of the device itself, such as the implementation
of the work tip displacement in accordance with the loaded G-code (Fig. 5).
Fig. 5. Visualization of Delta robot in EON studio software.
Research experiments were also planned in which simulation of the shape of virtual
objects was tested [16].
By definition, the main task of the Delta robot as an active touch device was to
simulate the shape of virtual objects. Therefore, as part of research, an experiment was
carried out which simulated the shape of the upper surface of the virtual cuboid by the
final effector. The task of the person testing the simulation was to try to touch it. The
following were visualized in the prepared VR application (Fig. 6a):
– digital robot model,
– a block of a virtual rectangle of blue color.
In order to map the shape of a virtual cuboid (defining the working space of the
Delta type manipulator), the coordinates of all vertices of the cuboid had to be included
in the script limiting the position of the digital robot in space.
In practice, the four tops of the upper surface were important (due to the fact that
the end effector platform made movements parallel to the reference plane). Thanks to
this, the touch device could not move beyond the designated area. Attempting to touch
the upper surface of the cuboid involved a change in the user’s hand position, which in
Fig. 6. a-Graphical interface of the test application operator (virtual robot displacement control),
b-Contact with the end effector of the Delta robot on the physical stand.
turn was recorded by the tracking system (a marker was placed on the user’s hand, the
object was tracked in real time). The task of the actual manipulator was to set the
working tip in the position enabling contact (i.e. near the shadowed hand of the user).
In this way, the person testing the simulation had the impression of contact with a
virtual object (in practice it was a physical artifact imitating a flat surface of a digital
cuboid, placed on the effector, Fig. 6b). To further enhance this feeling, the VR
application uses a digital hand model whose current position was registered by the
tracking system. Data about the current position of the digital hand was used to control
the position of the working tip of the virtual robot [16].
For the purposes of visualization of angular displacements of the arms and linear
movement of the virtual device’s working tip, the necessary simulations of the kine-
matic constraints have been defined in the simulation project. The VR application also
implements the workpiece positioning algorithm (reverse kinematics). To keep the
behavior of the virtual device close to the real one, the operator of the test application
monitored the speed of its movement on an ongoing basis. Each time the robot changed
position, the angular coordinates of its arms were automatically calculated and the
motion animation was displayed. All logic functions have been programmed with the
support of classical techniques (scripts prepared in VBScript) and visual techniques
(defining connections between nodes).
4 Summary
Engineering applications of Virtual Reality, especially in the design and construction of

machines and in production engineering are a key issue for the authors of the article.
The presented examples of the use of a VR for creating interactive simulations of
machines and devices are usually a prelude to further research work, and Virtual
Reality as a set of modern technologies is an important element of the whole concept of
Industry 4.0 [18, 19]. To sum up, virtual techniques play their role, as mentioned
earlier, at various stages of the product life cycle. The most important applications are
product design, process virtualization and digitally supported workstation training.
Acknowledgments. The research work reported here was partially financed by Polish Ministry
of Science and Higher Education.
References
1. Pająk, E., Dudziak, A., Górski, F., Wichniarek, R.: Additive manufacturing technologies and
virtual reality in processes of production preparation (2011). ISBN 978-83-86912-56-8. (in
Polish)
2. Górski, F., Hamrol, A., Grajewski, D., Zawadzki, P.: Integration of virtual reality and
additive manufacturing technologies – hybrid approach to product development. Mechanik
(3), 173–176 (2013). (in Polish). Cz. 2, Mechanik (4), 266–270
3. Brooks Jr., F.: Walkthrough - a dynamic graphics system for simulating virtual buildings. In:
Proceedings SIGGRAPH Workshop on Interactive 3D Graphics (1986)
4. Whyte, J., Bouchlaghem, N., Thorpe, A., McCaffer, R.: From CAD to Virtual Reality:
modelling approaches, data exchange and interactive 3D building design tools. Autom.
Constr. 10(1), 43–55 (2000)
5. Weiss, Z., Kasica, M., Kowalski, M.: Virtual reality in machine design. MACH-TOOL
2005, “Innowacyjne technologie w budowie maszyn” (2005). (in Polish)
6. Maleshkov, S., Katicic, J., Stojanova, P., Bachvarov, A.G.: Design-by-the-customer through
virtual reality. In: Advanced Research in Virtual and Rapid Prototyping, Proceedings of
VR@P4 (2009)
7. Weiss, Z., Konieczny, R., Diakun, J., Grajewski, D., Kowalski, M.: Integrated design at
VIDA centre. In: 19th CIRP Design Conference Cranfield, pp. 40–45 (2009). ISBN 978-0-
9557436-4-1
8. Grajewski, D., Górski, F., Hamrol, A., Zawadzki, P.: Immersive and haptic educational
simulations of assembly workplace conditions. Procedia Comput. Sci. 5, 359–368 (2015).
https://doi.org/10.1016/j.procs.2015.12.258
9. Górski, F., Buń, P., Wichniarek, R., Zawadzki, P., Hamrol, A.: Design and implementation
of a complex virtual reality system for product design with active participation of end user.
In: Advances in Intelligent Systems and Computing, vol. 492, pp. 31–43 (2016). https://doi.
org/10.1007/978-3-319-41935-0_4
10. Pandilov, Z., Milecki, A., Nowak, A., Górski, F., Grajewski, D., Ciglar, D., Mulc, T., Klaic,
M.: Virtual modelling and simulation of a CNC machine feed drive system. Trans. Famena
39(4), 37–54 (2016)
11. Cecil, J., Kanchanapiboon, A.: Virtual engineering approaches in product and process
design. Int. J. Adv. Manuf. Technol. 31(9–10), 846–856 (2007)
12. Ryan, R.: Digital Testing in the Context of Digital Engineering - Functional Virtual
Prototyping. Mechanical Dynamics Inc., Michigan (1999)
13. Trentin, A., Perin, E., Forza, C.: Product configurator impact on product quality. Int. J. Prod.
Econ. 135(2), 850–859 (2012)
14. Trojanowska, J., Karwasz, A., Machado, J., Varela, M.L.R.: Virtual reality based ecodesign.
In: Golinska-Dawson, P., Kolinski, A. (eds.) Efficiency in Sustainable Supply Chain, Part II,
pp. 119–135. Springer (2017). https://doi.org/10.1007/978-3-319-46451-0_8
15. Jimeno, A., Puerta, A.: State of the art of the virtual reality applied to design and
manufacturing processes. Int. J. Adv. Manuf. Technol. 33(9–10), 866–874 (2007). https://
doi.org/10.1007/s00170-006-0534-2
16. Grajewski, D.: Badanie interakcji dotykowej w rzeczywistości wirtualnej z zastosowaniem
robota typu delta, rozprawa doktorska (2018)
17. Ong, S.K., Nee, A.Y.C.: Virtual and Augmented Reality Applications in Manufacturing.
Springer (2004)
18. Brettel, M., Friederichsen, N., Keller, M., Rosenberg, M.: How virtualization, decentral-
ization and network building change the manufacturing landscape: an industry 4.0
perspective. Int. J. Mech. Aerosp. Ind. Mechatron. Manuf. Eng. 8(1), 37–44 (2014)
19. Żywicki, K., Zawadzki, P., Górski, F.: Virtual reality production training system in the scope
of intelligent factory. In: Advances in Intelligent Systems and Computing, vol. 637, pp. 450–
458. Springer (2017)
Cyber Sickness in Industrial Virtual
Reality Training
Magdalena Żukowska(&), Paweł Buń, Filip Górski,

and Beata Starzyńska

magdalena.k.zukowska@doctorate.put.poznan.pl
Abstract. The paper presents results of experimental studies of how cyber

sickness (also known as the simulator sickness) can influence basic tasks per-
formed by users of industrial VR (virtual reality) training simulations. Two
application cases are presented – a simple walk around a virtual environment
using two different movement possibilities and a more complex training process
with object interaction involved. They were tested with representatives of real
target groups – engineers and employees of production companies. The used
equipment involved mobile and stationary headsets. The obtained results con-
firm initial expectations about stationary headsets being generally more safe for
use regarding the cyber sickness issue. It was also found that continuous arti-
ficially induced movement in a VR application is generally less comfortable for
the users and therefore unadvised to employ in training scenarios.
Keywords: Virtual reality Manufacturing Industrial training

Cyber sickness
1 Introduction
Immersive virtual reality (VR) has been known for decades now, but has gained
considerable recognition in the beginning of this century. Industrial applications of VR
are very wide and still growing, especially in the concept of Industry 4.0 [1]. One of the
most prevalent professional VR application is education and training [2]. The positive
educational effect of VR has been observed and reported in literature as early as the
90’s [3] and proven in further studies [4, 5]. Today, VR is commonly used in engi-
neering education [6, 7], as well as in many other engineering activities [8] e.g. con-
figuration of variant products in the automotive branch [9, 10]. The largest automotive
or aeronautical manufacturers use VR for training purposes [11, 12] and this is mostly
due to continuous increase in availability of commercial solutions, marked by 2016
launch of considerably inexpensive but well-parameterized VR headsets, such as
Oculus Rift and HTC Vive. There has been also an increased demand for mobile
systems, with headsets based on popular smartphones [13]. As of second decade of
XXI. century, even small and medium enterprises can afford a VR system. Also,

https://doi.org/10.1007/978-3-030-18715-6_12
138 M. Żukowska et al.
another class of consumer VR systems emerged – represented by devices such as

HoloLens or Pimax 8K – very capable devices but more expensive than typical con-
sumer headsets.
A VR environment can heavily influence behavior and feelings of a user and can be
used for teaching knowledge and skills – this is mostly due to immersion, a unique
feeling induced by being cut off sensory signals from the real world [14]. However,
sometimes it can lead to a set of uncomfortable sensations generated by a brain,
especially when visual signals do not match what is experienced by a body of a user.
This is known as the cyber sickness (CS) and it has been always a large hindrance in
both professional and recreational widespread use of VR. The nature of this problem is
similar to Motion Sickness and generate similar symptoms, but are caused by different
environment – virtual (CS) and real (MS) world.
The most common symptoms of the sickness are: headache, nausea, vomiting, eyes
fatigue, postural instability, disorientation etc. It is a natural physiological reaction due
to mismatch between sensory inputs. In Motion Sickness there are differences between
signals send during movement – muscles of inner ear send information about it, but
person cannot see the changes. In Cyber Sickness the situation is quite the opposite – a
person can see their movement in virtual world, but cannot feel it, because natural
human motion is limited (cables, small tracking areas). This phenomenon based on
sensory conflict theory, where “passive movement creates a mismatch between infor-
mation relating to orientation and movement supplied by the visual and the vestibular
systems, and it is this mismatch that induces feelings of nausea” [15]. In addition, the
problem is individual, each person experiences CS in a different way, various envi-
ronments can cause divergent feeling in users. Some studies suggest that being in a
virtual environment 20 min can increase CS symptoms in over 60% of users [16]. Also
other studies shows that exposure to virtual reality caused eyestrain in 40% exposures,
nausea in 30% and other symptoms - headache, dizziness, losing postural stability or
sweating – to a lesser extent [17]. Cyber sickness is difficult to predict because of
individual standpoints [18]. There is also a problem in estimating long-term effects of
CS in a human body. That is why studies allow for different scenarios with diverse
activity of participants and with use of various type of hardware for VR.
Aim of the studies presented in this paper was to verify how CS can influence
activities performed during an industrial training. Groups of participants were subjected
to presence in two different VR applications, launched on two different headsets. Some
variability in interaction was also implemented.
2.1 Research Aims and Outline

The general aim of the studies on cyber sickness performed by the authors – which this
paper is a part of – is to answer a defining question on VR training. The question is – is
VR adequate for everyone in terms of training of manufacturing processes? This is a
very important problem – as VR was proven to be very efficient in industrial training
and obtaining between 75 and 90% time yield is possible [12, 19], it is justified to
Cyber Sickness in Industrial Virtual Reality Training 139
consider discarding traditional training procedures at all. However, if CS is an inherent

thing of modern VR that cannot be limited in training scenarios, then the training
cannot be purely VR-based, as certain individuals will be unable to complete it.
The studies in this paper focus on three important aspects:
1. Stationary versus mobile headsets – standard commercial devices were used in
similar conditions and scenarios, to find out which one induces more severe CS
symptomps.
2. Continuous versus instant point-to-point movement in virtual space – two methods
of movement were tested, walking with use of joystick and instant movement
(teleporting).
3. Simple versus complex interaction scenario – two different applications were pre-
pared, one involving a simple walk and the other – more complex training scenario
in a virtual factory.
Tests focused on methods of motion in virtual space, interactions with this world
and how it affects users of various age, gender and education. For evaluation of results,
a standard simulator sickness questionnaire was used by Kennedy et al. [20].
2.2 Hardware and Software

Two state-of-the art commercially available consumer headsets were used in the
studies. The stationary one was the Oculus Rift CV1 (Fig. 1a). The mobile headset was
the Samsung GearVR, paired with a Samsung Galaxy S6 mobile phone (Fig. 1b). The
main qualitative differences between the headsets – aside resolution, field of view
(FOV) and minor frame construction differences – are number of displays (two separate
displays in Rift, one in GearVR) and positional tracking of user’s head (mobile
headsets do not have this possibility).
In terms of software, both applications were created using the Unity 3D engine.
Originally, the applications were prepared for the PC/Oculus deployment. After pro-
gramming all the interactions, alternative versions of the software projects were made
and the platform was switched to Android/GearVR, with all the necessary adjustments
in graphics quality and input, not changing anything in interaction mechanics.
2.3 Experiment One – Virtual Walk

A simple virtual walk around a house, garage and workshop was arranged (see Fig. 2).
Two different types of movements were applied, as mentioned before:
1. With the use of teleports – special points in space; entrance in interaction allow for
transfer to different point in environment
2. Free walk with the use of a joystick.
The joysticks used in two headsets were different devices, but had the same form
factor and functionality (two-handed wireless gamepad with analog stick used for
moving). It is noteworthy, that the users of Oculus Rift headset had possibility to use
positional tracking and perform natural body movements in a space of 2 2 m. Users
Fig. 1. Devices used in the studies, (a) Oculus Rift CV1 [21], (b) Samsung GearVR [22]
Fig. 2. Experiment one – visualization of environment
of the GearVR did not have this possibility. Each test subject was free to assume a
posture of their liking (seated or standing).
Tests has been performed on group of 77 engineering students (mechanical engi-
neering, production engineering) aged 19–23, having not used VR technology before.
The participants had about 3 min to get familiarized with the hardware and the virtual
environment. Their assignment was to go around courtyard five times and then to the
north end of the building (Fig. 3). After the task, the participants could go around the
whole virtual environment, but maximum time of use could not exceed 5 min. Between
the different attempts, they had to take a rest for min. 10 min.
Fig. 3. Experiment one – walking path with marked teleporters and a mandatory location
2.4 Experiment Two – Virtual Quality Management

The second application is the “Virtual Quality Toolbox”, developed as a part of an
earlier project [23]. The Virtual Quality Toolbox is a VR solution that allows
employees of small and micro production companies to easily gain knowledge about
application of quality management tools in practice, by use of immersive VR [20].
The system prototype contains 7 quality tools (of several dozen existing), such as
Pareto diagram, Shewhart control chart, fishbone diagram, affinity diagram, among
others. Learning of a selected tool requires users to perform some activities in the
production hall, such as listening to the master, bringing a certain object or performing
a set of actions with it (Fig. 4).
In the tests, 11 participants aged 25–50 were involved. Their task was to complete a
single lesson on Shewhart control chart (expected time for completion was approx. 15–
20 min). Movement and interaction was realized using a gamepad and natural head
movements. The lesson involved a lot of object manipulation, as well as measurement
of products on a production line and activities such as listening to the production
master. The user was free to select a form of point-to-point transportation – teleports
were available as well as continuous move, it was a dynamic choice of a given
participant.
Fig. 4. Virtual Quality Toolbox – most important elements of the training [21]
3.1 Experiment One

The results of the experiment 1 are presented in Tables 1 and 2.
Symptoms can be grouped in three types:
• Oculomotor (headache, eyestrain, difficulty concentrating, difficulty focusing,
blurred vision) – see Fig. 5 for diagram of results,
• Disorientation (dizziness, vertigo) – see Fig. 6,
• Nausea (nausea, stomach awareness, increase of salivation, burping) – see Fig. 7.
As clearly visible in the diagrams, the most frequently occurring symptoms are
those related to disorientation and oculomotor types. The nausea group of symptoms
was reported generally less often and usually by the same participants. Although the
severity of symptoms varied, a general rule was that the persons who had nausea in the
Gear VR teleporting scenario, had it also in the Oculus Rift with the same movement
type, which is generally the safest (least CS-inducing) scenario out of all the 4 tested in
the experiment. Therefore, it can be said that digestive system symptoms of CS are rare
and generally based on personal characteristics, rather than induced by software means.
In terms of linking scenarios with CS symptoms – negative feelings and general
discomfort was pointed out the most often during free walk with gamepad, using the
Gear VR headset. It is probably caused by low framerate of the application, caused by
low computing power of a smartphone. Better performance of the Oculus Rift CV1
used with a well-configured PC increased comfort. No severe CS symptoms were noted
by any participant wearing this headset, regardless of movement method.
The disorientation symptoms were the most often in the gamepad movement sce-
nario as well and they decreased when the movement method was switched to instant
teleportation. That could be counterintuitive, as teleportation over large distances is
most certainly not an intuitive movement strategy for humans. However, the gamepad
creates a so-called sensor conflict – incompatibility between visual clues and body
awareness (inner ear signals, or lack thereof), as a virtual character moves without the
Table 1. Cyber sickness results – experiment 1, free walk; (a) GearVR, (b) Oculus Rift CV1
(a)
Symptoms GearVR
None Slight Moderate Severe
General discomfort 18% 75% 0% 6%
Fatigue 79% 14% 0% 6%
Headache 90% 9% 0% 1%
Eyestrain 71% 9% 19% 0%
Difficulty focusing 26% 61% 13% 0%
Salivation increasing 94% 6% 0% 0%
Sweating 97% 3% 0% 0%
Nausea 66% 27% 5% 1%
Difficulty concentrating 58% 42% 0% 0%
“Fullness of head” 66% 34% 0% 0%
Blurred vision 52% 48% 0% 0%
Dizziness with eye open 44% 43% 13% 0%
Dizziness with eye closed 55% 40% 17% 0%
Vertigo 31% 69% 0% 0%
Stomach awareness 74% 26% 0% 0%
Burping 100% 0% 0% 0%
(b)
Symptoms Oculus Rift CV1
Fatigue 87% 13% 0% 0%
Headache 96% 4% 0% 0%
Eyestrain 82% 9% 9% 0%
Sweating 100% 0% 0% 0%
Nausea 79% 21% 0% 0%
Blurred Vision 57% 43% 0% 0%
Vertigo 38% 62% 0% 0%
Burping 100% 0% 0% 0%
Table 2. Cyber sickness results – experiment 1, teleports; (a) GearVR, (b) Oculus Rift CV1
(a)
Symptoms GearVR
Fatigue 91% 9% 0% 0%
Headache 94% 6% 0% 0%
Eyestrain 86% 14% 0% 0%
Sweating 94% 6% 0% 0%
Nausea 71% 29% 0% 0%
Blurred vision 60% 40% 0% 0%
Vertigo 64% 36% 0% 0%
Burping 100% 0% 0% 0%
(b)
Symptoms Oculus Rift CV1
Fatigue 92% 8% 0% 0%
Headache 90% 9% 0% 0%
Eyestrain 91% 9% 0% 0%
Sweating 97% 3% 0% 0%
Nausea 75% 25% 0% 0%
Blurred Vision 65% 35% 0% 0%
Vertigo 71% 29% 0% 0%
Burping 100% 0% 0% 0%
Fig. 5. Oculomotor symptoms of cyber sickness in experiment 1
Fig. 6. Disorientation symptoms of cyber sickness in experiment 1
human body feeling the movement. That does not occur when teleporting, as sudden
changes of environment are not so intensive for a human being.
To sum it up, the most efficient (regarding time of operation) and least CS-inducing
scenario is combining the Oculus Rift headset with a teleport-based movement,
allowing users to walk a short distance (up to 2 m) around a teleporting spot – in the
majority of participants no symptoms occurred at all, with only occasional indication of
some symptoms as “moderate”, with majority of reported (other than “none”) symp-
toms marked as “slight”. This approach, however safe, has certain application design
drawbacks – the movement is not natural and the teleporters have to be available
throughout all the working space, which can be huge. Certain aspects of work during
training scenarios cannot be measured this way, such as times spent by employees
Fig. 7. Nausea symptoms of cyber sickness in experiment 1
walking from one workplace to another. Still, it should be always employed in

applications where long distance needs to be covered by a trainee, as an alternative,
safe solution.
3.2 Experiment Two

The results of the experiment 2 are presented in Table 3 and 4.
Table 3. Cyber sickness results – experiment 2, stationary headset

Oculus Rift CV1
Symptoms None Slight Moderate Severe
1 General discomfort 45% 55% 0% 0%
2 Fatigue 82% 9% 9% 0%
3 Headache 91% 9% 0% 0%
4 Eyestrain 36% 45% 18% 0%
5 Difficulty focusing 36% 55% 9% 0%
6 Salivation increasing 100% 0% 0% 0%
7 Sweating 100% 0% 0% 0%
8 Nausea 82% 18% 0% 0%
9 Difficulty concentrating 73% 27% 0% 0%
10 “Fullness of head” 55% 36% 9% 0%
11 Blurred vision 82% 18% 0% 0%
12 Dizziness with eye open 82% 18% 0% 0%
13 Dizziness with eye closed 73% 27% 0% 0%
14 Vertigo 82% 18% 0% 0%
15 Stomach awareness 91% 9% 0% 0%
16 Burping 91% 9% 0% 0%
Table 4. Cyber sickness results – experiment 2, mobile headset

Samsung GearVR
Symptoms None Slight Moderate Severe
1 General discomfort 18% 64% 9% 9%
2 Fatigue 55% 36% 9% 0%
3 Headache 64% 27% 9% 0%
4 Eyestrain 9% 18% 45% 27%
5 Difficulty focusing 18% 36% 27% 18%
6 Salivation increasing 91% 9% 0% 0%
7 Sweating 91% 9% 0% 0%
8 Nausea 64% 27% 9% 0%
9 Difficulty concentrating 27% 64% 9% 0%
10 “Fullness of head” 36% 45% 18% 0%
11 Blurred vision 27% 36% 27% 9%
12 Dizziness with eye open 73% 27% 0% 0%
13 Dizziness with eye closed 82% 18% 0% 0%
14 Vertigo 45% 45% 9% 0%
15 Stomach awareness 82% 9% 9% 0%
16 Burping 91% 0% 9% 0%
The results of the second experiment tend to confirm the results from the experi-
ment one. Again no severe symptoms were reported in the Oculus CV1 scenario.
Simultaneously, use of Gear VR caused some persons to not finish the lesson, because
the symptoms occurred quickly after commencing the test and were very uncomfort-
able. In general, the results are slightly worse than in the experiment one. This can be
attributed to three aspects:
• longer time spent in VR on average,
• higher age of participants – literature confirms that in VR older users appear to
suffer from CS more often than younger people [18],
• more complex interaction scenarios, requiring movement of user’s head in various
directions rapidly.
In both scenarios, general discomfort and difficulty focusing were the most often
reported problems. What is interesting in comparison with the experiment one, in both
test there were no major problems with dizziness. This can be attributed to the virtual
environment itself – in the experiment two, the users were located inside a virtual
building (interior), while in the experiment one, it was an open space with a sky and a
horizon. Another observation is that the participants, when given freedom of choice
regarding the movement strategy, preferred teleporting only over larger distances,
while in close spaces they preferred movement using a gamepad. This indicates that
cyber sickness related to movement is more prominent in the open space, as probably
the human sense of depth is more stimulated by the acquired visuals. The users feel
much more confident in smaller, closed spaces and they induce the CS the whole lower
lot than the large spaces. This has to be taken into account while creating virtual
environments and programming navigation.
4 Conclusions
Dynamic development of Virtual Reality and equipments lead to emergence new low-
cost products and it is associated with opening new areas in VR like inexpensive
training seats to learn basics in controlled environment. Unfortunately some interaction
in virtual world can generated severe problems with human body, concentration and
orientation known as Cyber Sickness. It could be potential barrier in implementation
trainings for everyone in factories and other fields (for example medicine).
The presented studies shown how important is movement in VR, constant free
motion could cause symptoms of CS and increase discomfort. It was found out that the
mobile VR headsets are – at the moment – no good for any professional training
whatsoever, as severity of CS symptoms greatly reduces effectiveness of knowledge
transfer. Probably in the future, when smartphones with greater performance will be
available, this can be reconsidered. Another very important observation – on the
positive side – is that using a commercially available VR headset paired with a rea-
sonably well configured PC it is possible to obtain zero reports on severe CS symp-
toms, regardless of complexity of interaction and time spent inside the application. This
is under a condition of allowing users to teleport around environment, without
disorientation-inducing free movement using a gamepad or any kind of analogue
device.
Future studies will involve larger scale tests on velocity of movement, more effi-
cient teleporting interaction and observing CS symptoms during longer sessions. It is
planned to perform these experiments in virtual factory, virtual power plant and virtual
garage, throughout regular training sessions in real production companies.
Acknowledgments. The research work reported here was made possible partially by Opera-
tional Program Knowledge Education Development, realized by Polish Ministry of Development
and partially be research grants of Polish Ministry of Science and Higher Education.
References
factory of the future. Int. J. Comput. Integr. Manufact. 30(1), 182–190 (2017)
2. Martín-Gutiérrez, J., Mora, E., Añorbe-Díaz, B., González-Marrero, A.: Virtual technologies
trends in education. EURASIA J. Math. Sci. Technol. Educ. 13(2), 469–486 (2017)
3. Bell, J.T., Fogler, H.S.: The investigation and application of virtual reality as an educational
tool. In: 1995 Annual Conference on Proceedings of the American Society for Engineering
Education (1995)
4. Blümel, E., Termath, W., Haase, T.: Virtual reality platforms for education and training in
industry. Int. J. Adv. Corp. Learn. 2(2) ( 2009). https://doi.org/10.1007/978-3-642-12082-4_1
5. Getchell, K., Miller, A., Nicoll, J.R., Sweetman, R.J., Allison, C.: Games methodologies and
immersive environments for virtual fieldwork. IEEE Trans. Learn. Technol. 3(4), 281–293
(2010). https://doi.org/10.1109/TLT.2010.25
6. Abulrub, A.H., Attridge, A., Williams, M.A.: Virtual reality in engineering education: the
future of creative learning. Int. J. Emerg. Technol. Learn. 6(4), 4–11 (2011)
7. Grajewski, D., Górski, F., Hamrol, A., Zawadzki, P.: Immersive and haptic educational
simulations of assembly workplace conditions. Procedia Comput. Sci. 75, 359–368 (2015).
https://doi.org/10.1016/j.procs.2015.12.258
8. Choi, S.S., Jung, K., Noh, S.D.: Virtual reality applications in manufacturing industries: past
research, present findings, and future directions. Concurr. Eng.-Res. Appl. 23(1), 40–63
(2015)
9. Górski, F., Buń, P., Wichniarek, R., Zawadzki, P., Hamrol, A.: Immersive city bus
configuration system for marketing and sales education. Procedia Comput. Sci. 75, 137–146
(2015). https://doi.org/10.1016/j.procs.2015.12.230
10. Jiang, M.: Virtual reality boosting automotive development. In: Ma, D., Gausemeier, J., Fan,
X., Grafe, M. (eds.) Virtual Reality & Augmented Reality in Industry, pp. 171–180.
Springer, Heidelberg (2011). https://doi.org/10.1007/978-3-642-17376-9
11. Hayden, S.: Volkswagen group to train 10,000 employees in VR this year, road to VR
(2018). https://www.roadtovr.com/volkswagen-group-to-train-10000-employees-in-vr-in-
2018/
12. Kishore, A.: Boeing: productive VR cuts training time by 75% (2017). lightreading.com.
Access 12 July 2018
13. Buń, P., Górski, F., Wichniarek, R., Kuczko, W., Hamrol, A., Zawadzki, P.: Application of
professional and low-cost head mounted devices in immersive educational application.
Procedia Comput. Sci. 75, 137–146 (2015). https://doi.org/10.1016/j.procs.2015.12.235
14. Bowman, D.A., McMahan, R.P.: Virtual reality: how much immersion is enough? Computer
40(7), 36–43 (2007). https://doi.org/10.1109/MC.2007.257
15. Colman, A.M.: A Dictionary of Psychology, 3rd edn. Oxford University Press (2014). http://
www.oxfordreference.com/view/10.1093/oi/authority.20110803100454911
16. Regan, E., Price, K.: The frequency and occurrence and severity of side-effects of immersion
in virtual reality. Aviat. Space Environ. Med. 65, 527–530 (1994)
17. Kennedy, R.S., Fowlkes, J.E., Lilienthal, M.G.: Postural and performance changes in navy
and marine corps pilots following flight simulators. Aviat. Space Environ. Med. 64, 912–920
(1993)
18. Martirosov, S., Kopecek, P.: Cyber sickness in virtual reality - literature review. In:
Proceedings of the 28th DAAAM International Symposium Cyber Sickness in Virtual
Reality - Literature Review, Vienna, Austria, pp. 0718–0726 (2017)
19. Górski, F., Zawadzki, P., Buń, P.: Virtual skill teacher – platform for effective learning of
technical skills. In: Proceedings of the International Conference of the Virtual and
Augmented Reality in Education (2018)
20. Kennedy, R.S., Lane, N.E., Berbaum, K.S., Lilienthal, M.G.: Simulator sickness question-
naire: an enhanced method for quantifying simulator sickness. Int. J. Aviat. Psychol. 3(3),
203–220 (1993)
21. https://pt.wikipedia.org/wiki/Oculus_Rift#/media/File:Oculus-Rift-CV1-Headset-Front_
with_transparent_background.png. Accessed 30 Sept 2018
22. https://www.skapiec.pl/site/cat/222/comp/13560785. Accessed 30 Sept 2018
23. Starzyńska, B., Górski, F., Buń, P.: Virtual reality system for learning and selection of
quality management tools. In: 2018 Advances in Intelligent Systems and Computing,
WorldCIST 2018, vol. 747. Springer (2018). https://doi.org/10.1007/978-3-319-77700-9_27
Does Metal Additive Manufacturing
in Industry 4.0 Reinforce the Role
of Substractive Machining?
Marcin A. Królikowski and Marta B. Krawczyk(&)
Faculty of Mechanical Engineering and Mechatronics,

Institute of Manufacturing Engineering,
West Pomeranian University of Technology, Al. Piastów 19,
70-310 Szczecin, Poland
mkrawczyk@zut.edu.pl
Abstract. The article describes in a review and debatable way a trend that can
be seen in modern industry, also known as the fourth industrial revolution or
Industry 4.0. After years of dynamic development of additive techniques from
polymeric, composite, ceramic and metal materials, it can be noticed that these
techniques have penetrated into the modern production reality. In particular, it is
used by such industries as space, air, automotive, precision, power, electronics
and the medical industry. The article focuses on the answer to the question
contained in the title. In almost all applications of metal additive manufacturing,
further processing is necessary in order to obtain the correct shape and/or
geometry accuracy. So there is a phenomenon of repeated interest in an accurate,
often multi-axis, substractive machining. The more problematic machining, the
more it is used in machining of elements with the so-called free forms resulting
even from optimizing the topology of additively manufactured metal elements.
The methods of metal additive manufacturing allow to obtain a geometry that
until now was impossible to achieve using conventional methods, and generates
additional technological problems in substractive machining. The answer to the
somewhat rhetorical question contained in the title of the article is self-evident,
and it is confirmed by observations from industry events and literature reports
related to classical and hybrid additive-substractive machining methods.
Keywords: Additive manufacturing Industry 4.0 Hybrid machining

Additive - substractive machining
1 Introduction – Industry 4.0
The fourth industrial revolution, called Industry 4.0, aims to increase economic com-
petitiveness by integrating modern production techniques with new information tech-
nologies [1, 2]. The key factors that enable the fourth industrial revolution include,
among others Internet of Things (IoT), Cloud Computing with response dynamics at
the level of milliseconds, Big Data - analysis of large data sets on all aspects of product

https://doi.org/10.1007/978-3-030-18715-6_13
Does Metal Additive Manufacturing in Industry 4.0 151
development and production, advanced human-machine interfaces, additive manufac-

turing techniques used both for prototyping and for the execution of individual orders
(Fig. 1) [1, 3].
Industry 4.0 is based on the paradigm of serial, personalized production, as a result
of which non-traditional methods of obtaining products complying with individual
customer requirements should be developed. Therefore, AM can become a key tech-
nology for the production of non-standard products due to the ability to create
advanced objects with advanced attributes (new materials, shapes) [1]. Thanks to the
increased product quality, additive manufacturing is currently used in various indus-
tries, such as aviation, biomedicine and production [2, 4]. As a rapidly developing
technology for creating accurate and complex objects, it can offer a way to replace
conventional production techniques in the near future.
Fig. 1. Scheme of intelligent factories divided into areas and main issues [1]
According to the Wohlers report from 2017 [5], at the moment over 60% of devices
for additive manufacturing are used to create functional parts, spare parts and for
education purposes. He predicts that with the development of technology, in the next
years AM techniques participation will increase in the production of final parts and
even finished products. He assumes two scenarios according to which the level of
production performed using traditional methods and additive methods will be equal in
2040 or at the latest in 2060. In the first case, he assumes that in 5 years, investment
outlays for 3D printers will increase twice, and in 10 years investment outlays on
traditional production machines will decrease by 33%, while in the second case,
investment outlays for 3D printers and traditional manufacturing machines will remain
the same as today. And although on the one hand the above predictions are promising
for both the devices manufacturers and for industries that already use AM techniques
152 M. A. Królikowski and M. B. Krawczyk
(i.e. aviation, automotive or medical industry), they also carry a negative impact. If the
above predictions prove reliable, due to the savings generated by additive technologies,
by 2060 the value of global trade will decrease depending on the scenario by 20% in
2040, or by nearly 25% in 2060. The reason for this is primarily the reduction of import
of mass-produced goods in favour of local, low-volume production on demand (this
will lead to a significant reduction in production costs, logistics, forwarding, etc.).
2 Techniques of Metal Additive Manufacturing (MAM)
Additive manufacturing techniques use a wide range of materials - from liquid crystal
and photocurable polymeric resins, polymer powders through thermoplastics in various
forms to metals. And the latter ones have the greatest opportunities in mass production
[1, 2, 6, 7]. Metal Additive Manufacturing (MAM) is already the most frequently
chosen technique for obtaining complex metal elements in industry [8] in almost all
engineering areas [1, 2, 6, 7]. Among the MAM techniques, the most common are
Selective Laser Melting and Direct Laser Metal Sintering (SLM/DLMS) implemented
in the automotive industry and Electron Beam Melting (EBM). These methods use
metal powders on the basis of powder bed. Figure 2 shows an example of the optimal
arrangement of small elements in the working space of the SLM device for serial
production. Figure 3 shows elements with a larger height after completing the process.
Fig. 2. A series of elements arranged in the working space for the purpose of the largest
packing. SLM method. Renishaw brand. [own source - FORMNEXT 2018]
Fig. 3. Elements on the production platform. Not cut off. [own source - FORMNEXT 2018]
The SLM/DLMS methods allow the production of metal elements with virtually
unlimited geometry on a unit or production scale. An integral process of additive
manufacturing is the removal of obtained elements from work platforms. The most
often used for this are electrical discharge machining methods. At trade fairs, more and
more companies are offering comprehensive solutions in the field of powder recycling
and removal of components from work platforms after MAM, including the recovery of
work platforms.
Methods using metal powders are characterized by the complexity of the process,
the high cost of equipment and material. In addition, they also require a very expensive
infrastructure due to the strong carcinogenic effects of powders. This aspect is finally
being revealed even at official trade fairs. This applies not only to the process itself but
also to pre- and post-processing, i.e. preparation and drying of the material, and then
machining. In each of these processes it is necessary to comply with strict safety
regulations (including the risk of ignition). EOS uses costly closed systems based on
cassette modules, where the workspace is a cassette placed in the device and picked up
by robots. However, this system requires additional, significant expenditure on pre-
processing and postprocessing of cassettes, transport and placement by robots. In order
to gain the trust of industrial customers, the company has built even a demonstration
factory fully serviced by robots. This factory is primarily intended to show that MAM
processes using powders can be carried out maintenance-free, and therefore safely.
Therefore, it causes a trend to create and promote alternative methods. Since it is
already known that it is not true, as it was originally announced that MAM methods
will completely or almost completely remove the classical substractive machining,
there appeared methods for which material removal is simply the next stage of the
technological process of shaping parts. One of the innovative methods with high
volumetric efficiency is WAAM [9]/SMD [10] (Wire + Arc Additive Manufacturing/
Shaped Metal Deposition) or commercial AMLTEC™ (Additive Metal Layering
Technologies), reserved for AML Technologies from the USA [11], which rely on
multi-layer, precisely controlled arc welding [9, 12–14]. Among the WAAM tech-
niques we can distinguish gas metal arc welding (GMAW) based, gas tungsten arc
welding (GTAW) based, and plasma arc welding (PAW) based processes [9]. PAW
technology is characterized by a higher energy density than GTAW causing less
deformation of the weld and also generating smaller welds at higher welding speeds
[15]. WAAM methods are the subject of many studies due to their complexity. For
example, Williams et al. [16] studied the effect of dealing with residual stress strategies
on mechanical properties and elimination of defects (porosity) in girders made of Ti-
6Al-4V, aluminum wing ribs or steel wind tunnels. Ding et al. [17] investigated the
influence of different wire feeding strategies on the quality and accuracy of the parts
produced. Zhan et al. [18] noted that the WAAM methods should examine many
different aspects affecting the final result of manufacturing, among others quality and
efficiency of used materials, designing production strategies, modeling and monitoring
of processes as well as their development. These methods have recently been imple-
mented in the aviation industry due to the ability to produce very large components and
the ability to shape all weldable metals [9, 12]. In particular, the aviation industry has
accepted additive methods based on hardfacing, not accepting any (besides bonding
and composites) methods of joining metals either by soldering or by welding. For these
methods, material removal is a shaping process as for semi-finished products made by
plastic forming or foundry methods. However, this is also a MAM method charac-
terized by great freedom of shaping and high volumetric productivity. Figure 4 shows
the raw part (a) and the machined element (b) after the WAAM method.
Fig. 4. (a) an unprocessed element produced by the WAAM method; (b) element after milling
[own source - FORMNEXT 2018]
In comparison to substractive machining, the WAAM method can reduce the

production time by 40–60% and the material removal time by 15–20% depending on
the size of the part produced [19]. For example, the use of WAAM technology for the
production of aircraft chassis rib saved about 78% of raw material compared to tra-
ditional substractive machining [16].
One of the MAM methods also used in the process of combined technologies, as
Wang et al. [20] write in a review on the production of titanium alloys, is a Laser
Cladding Method (CLAD) also called LMD (Laser Metal Deposition). This method
allows to apply layers to already existing objects creating multimaterial hybrids or
coated elements. It also allows to regenerate worn parts and, in the case of complex and
expensive parts of machine elements, to modify existing geometry.
Szost et al. [21] compared the microstructure and residual stresses generated in
components made of Ti-6Al-4V produced by WAAM and CLAD techniques. They
noticed that these methods allow to obtain fully dense structures with few porosities in
the vicinity of the motherboard, differing in microstructure (CLAD - narrow wavy
columnar grains and large columnar grain spreading on almost all the wall height in the
WAAM specimen), characterized by occurring stresses (the occurrence of maximum
residual stresses in the wall-baseplate transition in both samples with the higher value
in WAAM sample was observed).
Another method of increasing popularity is the MFDM (Metal Fused Deposition
Modeling) method, referring to the well-known FDM method. It is applicable to small
dimensions but it belongs to methods that do not require any industrial infrastructure
(desktop devices). The company with the characteristic name Desktop Metal has
developed a solution in which the entire process takes place in the office premises. This
is a breakthrough in the metal additive manufacturing which until now required lab-
oratory conditions or industrial infrastructure. At least during the final forming of metal
parts in the process of e.g. sintering. Elements created on MFDM devices from Desktop
Metal are shown in Fig. 5a, and the complete set is shown in Fig. 5b.
Fig. 5. (a) elements made with MFDM; (b) a complete station of the company Desktop Metal.
[own source - FORMNEXT 2018]
There is a trend to incorporate the classic FDM method into MAM (and its
derivatives). The company BASF (Fig. 6) has implemented, on an industrial scale,
high-filled thermoplastic materials in the form of monofilaments with diameters of 1.75
or 3 mm intended for classic, also low-budget, FDM devices.
Fig. 6. (a) high-filled with steel filament 316L; (b) standard FDM test model after FDM and
after sintering - element made of steel 316L. BASF company. [own source - FORMNEXT 2018]
The elements are then sintered under controlled conditions, as a result of which a
metal object is obtained (Fig. 6b). And the matrix is melted/burned out. However, one
observes here a volume loss and a significant change in the dimensions of the final
object. Although it is becoming a viable alternative to expensive MAM powder
methods in some applications [22]. The disadvantage of this method is the need to
master the process of post-processing by sintering. And this makes the process more
complicated than just “printing” using the FDM method.
3 Substractive Machining as the Next Stage of the

Technological Process of Metal Parts After AM
As a rule, machining of the right part is also necessary to correct the inaccuracy of
additive methods [23, 24]. This treatment is treated as finishing (postprocessing),
especially for elements of cooperating parts (Fig. 7).
Fig. 7. Examples of exact elements produced MAM and machined with conventional material
removal methods. [own source - FORMNEXT 2018]
Fig. 8. Elements generated by MAM methods with prepared gripping elements for processing
on a double-spindle milling and turning center [24].
It is worth noting that under Industry 4.0, a creation of hybrid methods is expected,
which will combine various additive techniques with each other or lead to different
combinations of production processes going beyond the conventional MAM processes
to produce better products with increased surface quality, endurance strength, etc. [1].
Hybrid machining processes, in particular the combination of additive and substractive
processing, require an individual approach to part modeling already at the stage of 3D
design in CAD programs. Since one of the biggest problems is just homing and fixing
of elements, it is necessary to predict the technology of substractive machining. Here
specially designed machining bases are used, which are later removed [24, 25] (Fig. 8).
This approach to hybrid machining not only provides “any” mounting but also,
depending on the functions and construction bases, enables to achieve the assumed
dimensional and shape accuracy. In special cases, it is possible to use as base elements the
production boards, which in this approach during the machining processes act as pallets.
Of course, the possibility of machining the lower fragments is then limited. Figure 9
shows the hybrid additive-substractive machining cycle carried out on a series of ele-
ments using technological bases generated together with the element during MAM.
Fig. 9. (a) CAD model; (b) surpluses imposed; (c) orientation in the machining space; (d) ensuring
the access and exit of the tool and the design of the gripping elements; (e) MAM; (f) fastening in a
machine tool; (g) determination; (h) generating tool paths; (i) finished element [24].
MAM methods using metal powders are used on an industrial scale, especially
where surfaces are curved or channels are to be, for example, conformal, as in aircraft
constructions. The obtained machining accuracy, treated as postprocessing, depends on
its proper preparation already at the design stage of AM parts and processes. This was
confirmed by Flynn et al., who in his work [26] noted that the synergistic combination
of additive manufacturing processes and substractive machining within a single
workstation allows to take advantage of each of them, facilitates the production of
elements with high shape factor with the desired geometrical accuracy and surface
characteristics. Also Manogharan et al. [25] noted that it is possible to improve the
efficiency of the process through a hybrid process consisting of EBM and rapid CNC
machining. It is worth adding here that Simons [27] in his article considered the
possibility of elimination of substractive methods in favour of additive manufacturing
in the production of metal parts. As a result of the analysis of costs, benefits and
changes that must be introduced to improve the competitiveness of AM, he noticed that
in order to achieve cost competitiveness of the AM technique it would have to have a
performance of at least 500 cm3 per hour. He also stated that material costs are a
significant limitation in AM techniques, which, however, will decrease over time as a
result of the increasing use of AM techniques not only in special production. A separate
problem in the introduction of MAM methods to the industrial everyday life is the need
for subcontractors and cooperators to adapt. They need to adapt to new realities and
technological barriers related to additive manufacturing methods as elements of a
typical technological process [28].
4 Hybrid Techniques; Additive-Subtractive Hybrid

Manufacturing (DASH) [29]
The main recipients of additive techniques - the automotive and aviation industries
mainly use metal AM processes [30]. However, additively manufactured metal parts
usually have large residual stresses caused by the heat due to the nature of the process
itself. Therefore, currently manufactured parts are subjected to heat treatment and
normalized before further processing. In addition, a finishing treatment is normally
required to achieve the surface quality required for certain types and areas of products
(bearing seats, cooperating flanges, threaded holes, etc.). The machining is also required
to remove the base plates and/or support elements. However, the above operations create
additional difficulties, because in order to be able to effectively process an additively
manufactured part, you need to know its exact shape, and this part should be properly
positioned and fixed in the machining space, which is crucial for the accuracy of the
dimensionally shaped part. The most promising way to solve the above problems and
even wider industrialization of AM solutions is their combination with subtractive
manufacturing (SM) processes [31]. The combination of both techniques gives the
opportunity to eliminate weaknesses and strengthen the stronger of each methods. It
enables the production of complicated parts in a shorter time ensuring the appropriate
quality. Kapil et al. [31] using Hybrid Layered Manufacturing (HLM) investigated the
impact of process parameters such as cladding current, stepover and torch speed have
been optimized to achieve the desired thickness of the layers and spatter free cladding.
They found that producing a turbine blade as part of a hybrid process is 68.29% more
efficient compared to conventional techniques. Stavropoulos et al. in their article [30]
they proposed solutions for additive manufacturing aimed at solving existing problems
by: complete personalization of machines in all aspects (work space, efficiency, quality
of parts, monitoring and control system, etc.), use of existing equipment (laser source,
positioning systems, sensors, SM head, etc.) to create a hybrid AM-SM machine and the
possibility to re-design and change the machine or its components to meet new needs
and even the re-use of machine components in various applications (modular approach).
Kendrick et al. [32] considered the use of hybrid systems in local factories - Just-in-Time
production (JIT), in specialized stores, using Fab Lab and at home. They note that the
use of the proposed solutions will reduce transport costs and the costs of storing various
stocks also reducing production lines with JIT. Jones et al. [33] described the use of
hybrid techniques (Laser Cladding and CNC) for regeneration of used turbine blades.
They emphasized that a fully integrated approach to the regeneration process may
eliminate the need to configure a single operation on each machine at all stages of
manufacturing, which are not only laborious, but above all generate a series of errors
resulting from setting the machine tool to a specific machining operation and mounting.
This approach is used on the one hand to increase dimensional accuracy, and on the
other hand to accelerate the entire production process. Also the problem of manufac-
turing complicated parts, where a single production process (substractive or additive) is
insufficient, can be increased by means of hybrid techniques [34]. The combination of
selective laser melting and precise milling to achieve the desired surface finish was also
introduced by Du et al. [35].
A certain reversal of the issue is the use of a head using LMD/DMD technology
(Laser Metal Deposition/Direct Metal Deposition) as one of the tools used on metal
cutting machine tools. An integral part of the machining process is the additive man-
ufacturing and then machining (material removal) as postprocessing. It can be used as a
production process or a process of part regeneration or correction of errors at a con-
struction or technological level.
In April 2018 a report on the development of hybrid techniques was published [36].
According to the report, the development of the hybrid additive manufacturing
(HAM) machinery market is related to their wider use in heavy, automotive, aviation,
medical, energy and electronic industries. It distinguishes three branches of the use of
hybrid additive manufacturing techniques: repairs, production and prototyping. It has
been observed that the production of complex components is leaving the conventional
production methods in favor of additive techniques. This is caused by the specificity of
the processes themselves - the material is added layer by layer, instead of selecting
material from the blank for the purpose of mapping geometry as it is in conventional
techniques - e.g. milling. This results in shorter production times and significant sav-
ings in material consumption and leads to a reduction in component production costs. It
is assumed that the aviation segment will have the main market participation due to the
increasing speed of using hybrid techniques. He used HAM to produce critical parts
that are both light and durable. Fulfillment of these requirements at the highest level is
guaranteed by HAM machines. In Europe Airbus, the leading aerospace manufacturer,
works with the Swiss Oerlikon group for the production of structural titanium com-
ponents. According to [30], the car industry has 21% of the total number of additive
manufacturing devices. It is anticipated [36] that the European market will grow in a
significant CAGR over the forecast period. In February 2018, CECIMO, a Europe-
based association of the machine tool industry signed an agreement with EPMA,
European Powder Metallurgy Association for the promotion of additive manufacturing
and to support the industry in Europe from one application of 3D printing to the other.
The report also details the main players involved in the development of HAM:
DMG MORI Co., Ltd., Mazak Corporation, Stratasys Ltd, voxeljet AG, Optomec,
SLM SOLUTIONS GROUP AG, Matsuura Machinery Corporation and Renishaw Plc.
For example, Matsuura Machinery Corporation introduced the Matsuura LUMEX
Avance-60 (Fig. 10) to the market using Metal Laser Sintering, which in standard
mode can build parts in 3D at a speed of 36 cubic centimeters per hour.
Fig. 10. View of the workspace of the LUMEX device during the milling process of the element
without liquid cooling directly after the additive manufacturing process [own source -
FORMNEXT 2018]
Figure 11 shows the element produced and processed on the LUMEX device.
Fig. 11. Element produced and processed in the HAM process on the LUMEX device [own
source - FORMNEXT 2018]
DMG MORI introduced LASERTEC 65 3D hybrid using Laser Deposition

Welding and 5-axis simultaneous milling in one mount and LASERTEC 30 SLM
based on Selective Laser Melting technique. Figure 12 shows a bi-material element
with interesting geometry manufactures and processed on a device using Laser Clad-
ding technology, made by DMG MORI.
Fig. 12. Bi-material elements produced on DMG MORI LASERTEC 65 3D devices in the
hybrid manufacturing process [own source - FORMNEXT 2018]
Mazak Corporation has created an INTEGREX i-400AM machine that using Laser
Deposition Welding and 5-axis machining can combine different types of metals,
allowing efficient repair of existing worn or damaged components such as air turbine
blades. All the above-mentioned devices are used to produce complex components that
can not be produced using conventional production methods. Considering the ever
wider applications of hybrid techniques and the faster development of devices used in
them, according to the report, it is estimated that the global hybrid additive-
manufacturing machines market is expected to reach US$ 6,757.1 Mn by 2026. The
market is projected to expand at a CAGR of 21.69% during the forecast period. It
should be added that device manufacturers are currently investing in software that
allows integration of machining processes in integrated CAD/CAM systems, as has
been the case with subtractive machining processes for years. CAD/CAM software
creators already commercially offer design of technological processes using this type of
machining, calling it a hybrid machining (e.g. Siemens NX).
Summary
To conclude the discussion and review that were carried out earlier, the answer to the
question contained in the title should be affirmative. With the increase of the impor-
tance of industrial metal additive machining, the complexity of machining processes
increases considerably both as a separate element of the technological process of
manufactured parts as well as in the case of combined, additive - subtractive
machining. The most important machining tool makers have already adapted their
products to the changing market and not only does the hybrid machining begin to be
noticed, but above all the formation of hybrid machining tools. This is caused by
considerably lower machining costs, if it is carried out in one mounting on one
machine. What’s more, virtually all major producers of computer aided manufacturing
software already present products that support additive manufacturing or hybrid
machining. The increase in the importance of machining (material removal), in par-
ticular multiaxial machining, is dictated by the greater possibilities of producing
geometrically complex elements from hard-to-machine materials. Particularly SLS,
SLM, DLMS and LMD or Laser Cladding methods allow to manufacture elements of
special alloys and super alloys. While the progress in the design and manufacture of
modern tool materials and coatings allows for effective machining, the bar increases the
geometrical complexity that makes the cutting process run in difficult conditions. It is
also worth mentioning the growing popularity of WAAM methods generating with
high volumetric capacity objects at low level of shape reproduction. In this case, the
increase in the importance of cutting machining consists in the necessity of using high-
performance methods of multi-axis shaping and even multi-axis roughing. Interest-
ingly, the aviation industry, which does not accept joining metals by welding, very
quickly accepted the WAAM method, which is a method of arc welding. This generates
further challenges for high-performance machining processes.
Additive manufacturing processes are included as components of typical manu-
facturing processes and start to be operated by computer-aided manufacturing
(CAM) systems as typical technological processes combined with manufacturing
processes by subtractive machining methods. Similarly, they begin to be elements of
PPC (Production Planning and Control) and ERP (Enterprise Resource Planning)
systems. Slowly combining additive and subtractive machining will cease to be a
hybrid machining and will become a classic in the fourth industrial revolution that is
already taking place. The progress in the field of adaptation of the market of devices,
tools and infrastructure to the requirements of Industry 4.0 allows to conclude that it is
actually no longer a revolution. In industrialized regions - it is everyday life.
Acknowledgments. Authors want to thank the Activ Reha Med Company for support.
References
1. Dilberoglu, U.M., Gharehpapagh, B., Yaman, U., Dolen, M.: The role of additive
manufacturing in the era of industry 4.0. Procedia Manufact. 11, 545–554 (2017)
2. Królikowski, M.A., Krawczyk, M.B.: Metal cutting as an integral process of metal additive
manufacturing in industry 4.0 (in Polish). Mechanik (8–9), 769–771 (2018)
3. Liu, Y., Xu, X.: Industry 4.0 and cloud manufacturing: a comparative analysis. J. Manufact.
Sci. Eng. 139(3), 1–8 (2016)
4. Thompson, M.K., Moroni, G., Vaneker, T., Fadel, G., Campbell, R.I., Gibson, I., Bernard,
A., Schulz, J., Graf, P., Ahuja, B., Martina, F.: Design for additive manufacturing: trends,
opportunities, considerations, and constraints. CIRP Ann. Manufact. Technol. 65(2), 737–
760 (2016)
5. https://www.ingwb.com/insights/research/3d-printing-a-threat-to-global-trade. Access 16 Mar

2018
6. Huang, Y., Leu, M.C., Mazumder, J., Donmez, A.: Additive manufacturing: current state,
future potential, gaps and needs, and recommendations. J. Manufact. Sci. Eng. 137(1),
pp. 014501–014501-8 (2015)
7. Attaran, M.: The rise of 3-D printing: the advantages of additive manufacturing over
traditional manufacturing. Bus. Horiz. 60(5), 677–688 (2017)
8. Sames, W.J., List, F.A., Pannala, S., Dehoff, R.R., Babu, S.S.: The metallurgy and
processing science of metal additive manufacturing. Int. Mater. Rev. 61(5), 1–46 (2016)
9. Pan, Z., Ding, D., Wu, B., Cuiuri, D., Li, H., Norrish, J.: Arc welding processes for additive
manufacturing: a review. In: Chen, S., Zhang, Y., Feng, Z. (eds.) Transactions on Intelligent
Welding Manufacturing. Springer, Singapore (2017)
10. Yilmaz, O., Ugla, A.A.: Microstructure characterization. Int. J. Adv. Manufact. Technol. 89
(1–4), 13–25 (2017)
11. Cegielski, P., Skublewska, A., Gawroński, P., Ostrysz, M., Dylewski, M., Gajowniczek, M.:
Robotized 3D printing of the machines parts with welding methods. Weld. Technol. Rev. 89
(1), 35–42 (2017). (in Polish)
12. Wang, F., Williams, S., Colegrove, P., Antonysamy, A.A.: Microstructure and mechanical
properties of wire and arc additive manufactured Ti-6Al-4 V. Metall. Mater. Trans. A Phys.
Metall. Mater. Sci. 44(2), 968–977 (2013)
13. Szost, B.A., Terzi, S., Martina, F., Boisselier, D., Prytuliaka, A., Pirlin, T., Hofmanne, M.,
Jarvisa, D.J.: A comparative study of additive manufacturing techniques: residual stress and
microstructural analysis of CLAD and WAAM printed Ti–6Al–4 V components. Mater.
Des. 89, 559–567 (2016)
14. Frazier, W.E.: Metal additive manufacturing: a review. J. Mater. Eng. Perform. 23(6), 1917–
1928 (2014)
15. Mannion, B., Heinzman, J.: Plasma arc welding brings better control. Tooling Prod. 5, 29–
30 (1999)
16. Williams, S.W., Martina, F., Addison, A.C., Ding, J., Pardal, G., Colegrove, P.: Wire + arc
additive manufacturing. Mater. Sci. Technol. (U. K.) 32, 641–647 (2016)
17. Ding, D., Pan, Z., Cuiuri, D., Li, H.: Wire-feed additive manufacturing of metal components:
technologies, developments and future interests. Int. J. Adv. Manufact. Technol. 81(1–4),
465–481 (2015)
18. Zhan, Y.M., Chen, Y., Li, P., Male, A.T.: Weld deposition-based rapid prototyping: a
preliminary study. J. Mater. Process. Technol. 135(2–3), 347–357 (2003)
19. Wu, B., Pan, Z., Ding, D., Cuiuri, D., Li, H., Xu, J., Norrish, J.: A review of the wire arc
additive manufacturing of metals: properties, defects and quality improvement. J. Manufact.
Process. 35, 127–139 (2018)
20. Weng, F., Chen, Ch., Yu, H.: Research status of laser cladding on titanium and its alloys: a
review. Mater. Des. 58, 412–425 (2014)
21. Szost, B.A., Terzi, S., Martina, F., Boisselier, D., Prytuliak, A., Pirling, T., Hofmann, M.,
Jarvis, D.J.: A comparative study of additive manufacturing techniques: Residual stress and
microstructural analysis of CLAD and WAAM printed Ti–6Al–4V components. Mater. Des.
89, 559–567 (2016)
22. Gong, H., Snelling, D., Kardel, K., Carrano, A.L.: Comparison of stainless steel 316L parts
made by FDM- and SLM-based additive manufacturing processes. JOM: J. Miner. Metals
Mater. Soc. (2018). https://doi.org/10.1007/s11837-018-3207-3
23. Królikowski, M., Filipowicz, K.: Verification of geometrical accuracy of selective laser
melting (SLM) built model. Adv. Manufact. Sci. Technol. 37, 85–91 (2013)
24. Frank, M.C., Harrysson, O., Wysk, R.A., Chen, N., Srinivasan, H., Hou, G., Keough, C.:
Direct additive subtractive hybrid manufacturing (DASH) – an out of envelope method, solid
freeform fabrication. In: Proceedings of the 28th Annual International, Solid Freeform
Fabrication Symposium – An Additive Manufacturing Conference, Austin, TX (2017)
25. Manogharan, G., Wysk, R., Harrysson, O., Aman, R.: AIMS- a metal additive-hybrid
manufacturing system: system architecture and attributes. Procedia Manufact. 1, 273–286
(2015)
26. Flynn, J.M., Shokrani, A., Newman, S.T., Dhokia, V.: Hybrid additive and subtractive
machine tools – research and industrial developments. Int. J. Mach. Tools Manufact. 101,
79–101 (2016)
27. Simons, M.: Additive manufacturing—a revolution in progress? Insights from a multiple
case study. Int. J. Adv. Manufact. Technol. 96(1–4), 735–749 (2018)
28. Strong, D., Kay, M., Conner, B., Wakefield, T., Manogharan, G.: Hybrid manufacturing –
integrating traditional manufacturers with additive manufacturing (AM) supply chain. Addit.
Manufact. 21, 159–173 (2018)
29. Frank, M., Harrysson, O., Wysk, R., Chen, N., Srinivasan, H., Hou, G., Keough, C.: A
Method for Integrating Additive and Subtractive Operations for Metal Parts – Direct
Additive Subtractive Hybrid Manufacturing (DASH), Conference Tools for Materials
Science & Technology 2017
30. Stavropoulos, P., Foteinopoulos, P., Papacharalampopoulos, A., Bikas, H.: Addressing the
challenges for the industrial application of additive manufacturing: towards a hybrid
solution. Int. J. Lightweight Mater. Manufact. 1(3), 157–168 (2018)
31. Kapil, S., Legesse, F., Kumar, R., Karunakaran, K.P.: Hybrid layered manufacturing of
turbine blades. Mater. Today Proc. 4(8), 8837–8847 (2017)
32. Kendrick, B.A., Dhokia, V., Newman, S.T.: Strategies to realize decentralized manufacture
through hybrid manufacturing platforms. Robot. Comput.-Integr. Manufact. 43, 68–78
(2017)
33. Jones, J., McNutt, P., Tosi, R., Perry, C., Wimpenny, D.: Remanufacture of turbine blades
by laser cladding, machining and in-process scanning in a single machine. In: Proceedings
from the 23rd Annual International Solid Freeform Fabrication Symposium (2012)
34. Kerbrat, O., Mognol, P., Hascoet, J.-Y.: A new DFM approach to combine machining and
additive manufacturing. Comput. Ind. 62, 684–692 (2011)
35. Du, W., Bai, Q., Zhang, B.: A novel method for additive/subtractive hybrid manufacturing of
metallic parts. Procedia Manufact. 5, 1018–1030 (2016)
36. Raport: Hybrid additive manufacturing machines – global industry analysis, size, share,
growth, trends and forecast, 2018–2026. www.transparencymarketresearch.com/hybrid-
additive-manufacturing-machines-market.html. Access: 16 Sept 2018
Ensuring Vibration Reliability of Turbopump
Units Using Artificial Neural Networks
Ivan Pavlenko1(&), Vitalii Ivanov1, Ivan Kuric2, Oleksandr Gusak1,

and Oleksandr Liaposhchenko1
1
Sumy State University, 2 Rymskogo-Korsakova Street, Sumy 40007, Ukraine
i.pavlenko@omdm.sumdu.edu.ua
2
University of Zilina, 8215/1 Univerzitna Street, 010 26 Zilina, Slovakia
Abstract. This paper is devoted to developing the scientific approach of using

artificial neural networks for solving a significant problem of vibration reliability
of rotary machines that is urgently needed to improve the quality of their
diagnosis and manufacturing. The proposed methodology integrates analytical
dependencies, recent techniques of numerical simulations and artificial neural
networks. The design schemes for realizing the related approach are presented
on the example of the turbopump unit for liquid rocket engine. The main
advantage of this approach in comparison with the traditional regression analysis
and other existing techniques is absence of necessity for setting trial imbalances
and carrying out additional initial starts of the turbopump unit. The mathe-
matical model for identification nonlinear parameters of the dependence
between bearing stiffness, deflection of the rotary axis, and rotor speed is pre-
sented. The proposed methodology is proved by the research of rotor dynamics
on the example of turbopump units for liquid rocket engines and allows refining
parameters of the nonlinear mathematical models describing forced oscillations
of the rotor as a complicated mechanical system with nonlinearities. The results
of the research can be used for carrying out the virtual balancing procedure for
identification the system of imbalances by the reliable model of forced oscil-
lation of the system “rotor – bearing supports”.
Keywords: Reliability Bearing support Stiffness

Nonlinear characteristics Artificial neural network
1 Introduction
Rotary machines take the significant role in the up-to-date production. Especially,
recent turbopump units for aerospace industry are permanently designed for providing
their reliable work in overestimated parameters and high precision quality requirements
for their manufacture. Primarily, ensuring the vibration reliability of the rotary
machines is an urgent problem for identification of the vibration characteristics in the
conditions of designing for extremely increased parameters and operation in real
conditions.
Recently, a lot of research works are devoted to investigation of the stated problem
of ensuring vibration reliability of rotors. Particularly, the research work [1] is aimed at

https://doi.org/10.1007/978-3-030-18715-6_14
166 I. Pavlenko et al.
developing the condition-based maintenance approach for providing maintenance

decisions based on the condition monitoring dataset. It consists of several substeps
including data acquisition, processing and decision-making. The main aspects in
application of this program (diagnostics and prognostics) are indicated. Recent tech-
niques in diagnostics of mechanical systems are summarized, as well as the proposed
maintenance approach, its models and algorithms for data processing and decision-
making are implemented.
In the paper [2], the approach of the fault diagnosis for rotary machines is inves-
tigated using empirical mode decomposition as a comparatively new method of time-
frequency analyzing for nonstationary nonlinear signals. However, this research shows
that sometimes this approach does not allow revealing the signal characteristics due to
the huge impact of external noises. For solving this problem, the combined mode
function is proposed for increasing accuracy of the identification of fault characteristics
of the rotating machine.
The previous research [3] was aimed at investigating rotor dynamics of the mul-
tistage centrifugal oil pump with ball bearings using the proposed computer programs,
which realize the mathematical models of free and forced oscillations of the rotor
analytically using finite element method (FEM) considering the dependence of bearing
stiffness on rotor speed defined previously by the experimental results. In addition, in
the papers [4, 5] the theoretical foundations and practical approaches are proposed for
providing static and dynamic analysis of the pump’s rotor considering the automatic
balancing device and pressure difference regulator. The related static and dynamic
characteristics are determined, as well as critical frequencies and dynamic stability of
the comprehensive system “rotor – automatic balancing device – pressure difference
regulator” are investigated.
The research works [6, 7] are aimed at realizing numerical and experimental
investigations of centrifugal compressor stage using the aerodynamic stand and up-to-
date software. As a result, the velocity distribution is estimated numerically for the
different operating parameters, and related recommendations to increase the efficiency
have been proposed. Additionally, the paper [8] realizes the analysis of change of face
impulse seal rings for rotary machines, which are completely made of wear-resistant
materials. It has been established that the seal rings surfaces are not in contact per-
manently. This fact allows summarizing the vibration reliability of rotors depends on
parameters of seals. Moreover, the approach for the development of intelligent auto-
mated systems for lean industrial production ensuring maximum efficiency is presented
in the research papers [9, 10].
Nevertheless, despite the above mentioned traditional approaches for carrying out
research in the field of providing vibration reliability of rotors, there is a huge amount
of research works relating to the use of artificial neural networks (ANN). Particularly,
the papers [11–13] are aimed at using wavelet analysis based on artificial neural
networks (ANN) for the fault diagnostics of rolling bearings. Particularly, the proposed
procedure is illustrated different bearing faults using the experimental data for the
rotary machine. The results prove the effectiveness of the proposed classifiers for the
detection of the bearing conditions for different success rates. Recently, both of sta-
tistical methods to preprocess the vibration signals and artificial neural networks
(ANN) for identification their parameters are widely used to detect faults in rotary
Ensuring Vibration Reliability of Turbopump Units Using ANN 167
machines. However, it has been shown [14] that using ANN is more successful in
comparison with common approaches.
The periodic multi-component signals measured by accelerometers allow obtaining
rich information for the purposes of vibration diagnosis of bearings. However, existing
approaches are mostly based on long-term data preprocessing to accurately obtain the
fault characteristics. Due to these facts, the papers [15, 16] are aimed at developing the
effective methods of transformation of the vibration signal into the plane image. For
this purpose, authors proposed to use the convolutional neural network able to acquire
the most suitable features. Experimental results proved that the accuracy of proposed
approach exceed the common methods of vibration diagnosis.
Fundamentally, in the paper [17], the general scientific and methodological
approach for the identification of mathematical models of mechanical systems using
artificial neural networks (ANN) is proposed for solving complicated interdisciplinary
problems in the field of manufacturing, mechanical and chemical engineering. This
approach is based on the comprehensive implementation of the analytical methods of
the research, modern methods of numerical analysis, and ANN.
The research works [18, 19] emphasize that the prediction of the remaining useful
life for accurate equipment takes the highly significant role for improving reliability
and reducing overall maintenance cost. In this paper, the approach based on using ANN
is developed for identification remaining useful life prediction of equipment in terms of
monitoring the operating conditions. Particularly, the author’s methodology is vali-
dated for pump bearings using vibration monitoring data, as well as a comparison is
proposed between the common approach and method based on using ANN. As a result,
the advantage of the proposed ANN method is proved.
The research work [20–23], methods for the identification unbalances in rotor
bearing systems is proposed using ANN. The method contains two different approa-
ched. The first one operates statistical information and is used to test ANN by the
frequency domain. In the second case, the infrequency domain is used.
The ANN method for identification of bearing stiffness characteristics of rotary
systems by their critical speeds is presented in the paper [24, 25]. Particularly, the
improved mathematical model of free oscillations is proposed and realized by the finite
element method (FEM), as well as the related software is developed. Finally, ANN
method and the related computational approach are proposed for evaluating bearing
stiffness by the critical speeds of the rotor.
Despite the presence of a wide variety of methods for increasing vibration relia-
bility of rotary machines and equipment, there is no unified approach for identification
parameters by the experimental data considering compliance of housing parts, non-
linear dependence of bearing stiffness on the rotor speed, as well as gyroscopic
moments of inertia of impellers and shell-type parts. Moreover, existing methods need
to use the data of preloading rotor by the system of trial imbalances and carrying out
additional initial starts of the turbopump unit. Due to the abovementioned, the main
aim of the research is to propose and implement the method for identification of
nonlinear stiffness characteristics of bearing supports using ANN for ensuring vibra-
tional reliability of the equipment based on mathematical models of their forced
oscillations and realizing virtual balancing procedure.
2 Research Problem
Due to the FEM approach, forced oscillations of the rotor is described analytically by
the matrix equation
ð½C x2 ½MÞfYg ¼ fDgx2 ð1Þ
with the following parameters: {Y} – row-vector m 1 of axial deflections; {D} –

row-vector n 1 of imbalances; m, n – number of measuring and correction planes
correspondently; [C] – stiffness matrix; [M] – matrix of inertia; x – rotor speed.
The first equation can be rewritten in the factored form [26]:
½WfDg ¼ fYg ; ð2Þ
where [W] – rectangular matrix of weighting factors:
½W ¼ x2 ð½C x2 ½MÞ1 : ð3Þ
The problem of identification the system of imbalances is solved analytically by the

linear regression procedure:
fDg ¼ ðfWgT fWgÞ1 fWgT fYg : ð4Þ
However, this equation does not take into account the dependence between stiffness
matrix [C], axial deflections {Y} and rotor speed x. Particularly, it is sufficient to
practical purposes to consider the following dependence:
Cij\k [ ¼ c\0
ij
[
þ aij x2k þ bij jYj\k [ j ; ð5Þ
where i = 1, 2, …, m – index of measuring plane; j = 1, 2, …, n – index of correction

plane; k = 1, 2, …, p – index of rotor speed; p – number of rotor speeds; c\0 ij
[
–
elements of linear stiffness matrix in case of zero rotor speed and axial deflections; aij,
bij – coefficients needed to be identified.
If m < n, the number of rotor speeds must be varied by the number p until the
following inequality will be satisfied: mp n. Besides, the Eq. (1) cannot be solved
directly. However, the proposed virtual balancing procedure allows identifying
imbalances directly under the condition of sufficient reliability of the abovementioned
mathematical model. In this case, ANN can be appropriately applied as described
below. The related procedure for identification of parameters for rotor bearing stiffness
characteristic by combined using FEM, ANN and experimental research is schemati-
cally presented in Fig. 1.
Fig. 1. The design scheme of for identification of parameters for rotor bearing stiffness
characteristic by combined using FEM, ANN and experimental research.
The first step includes FEM analysis for calculation row-vector of deflection {Y} by
the set of imbalances {D} for different operating speed x. As a result of numerical
simulation, the input and output data are transferred to the output and input data of the
ANN architecture correspondently. After ANN training, the system of imbalances {D}
is evaluated, as well as training accuracy is calculated. Additionally, the experimental
dataset allows obtaining predicting the system of imbalances {D}. Finally, the com-
parison of results using ANN and experimental data allows estimating ANN accuracy.
3 Results
The following research of rotor dynamics is realized for the turbopump unit with the
design scheme presented in Fig. 2. The unit has p = 3 operating speeds: x1 = 1100
rad/s, x2 = 1963 rad/s, and x3 = 2215 rad/s. Previous investigations of bearing stiff-
ness [24] according the critical frequencies of the rotor allows determining the fol-
lowing parameters of the formula (5): c\0
1
[
= 1.88108 N/m; c\0
2
[
= 2.10108 N/m;
a1 = 1.223 N s /m; a2 = 0.408 N s /m; b1 = 0.1810 N/m ; b2 = 2.241012 N/m2.
2 2 12 2
Fig. 2. 3D model of the rotor and its design scheme.
Table 1. The results of numerical simulation

Imbalances, Rotor speed, rad/s Deflections, lm
kg mm
D1 D2 x Y1 Y2 Y3
1 0 1100 13.2 1.1 –1.3
2 0 1100 26.1 2.1 –2.5
3 0 1100 38.9 3.1 –3.7
0 1 1100 1.1 4.7 0.7
0 2 1100 2.2 9.4 1.4
0 3 1100 3.3 14.1 2.1
1 0 1963 64.2 8.8 –6.1
2 0 1963 121.3 15.8 –11.1
3 0 1963 173.4 21.6 –15.4
0 1 1963 9.6 19.8 2.9
0 2 1963 18.8 39.2 5.2
0 3 2215 27.4 58.3 7.0
1 0 2215 102.4 16.7 –9.3
2 0 2215 184.6 28.2 –16.2
3 0 2215 256.0 36.9 –21.8
0 1 2215 19.8 29.5 3.8
0 2 2215 37.4 57.8 6.5
0 3 2215 53.0 85.3 8.4
1 1 1100 14.2 5.8 –0.5
2 2 1100 28.1 11.5 –1.0
3 3 1100 41.7 17.1 –1.4
1 1 1963 71.6 28.0 –2.9
2 2 1963 133.0 53.3 –4.8
3 3 1963 187.8 76.8 –6.3
1 1 2215 115.1 44.3 –4.7
2 2 2215 202.4 81.0 –7.2
3 3 2215 275.7 113.7 –8.7
Due to the design scheme, it is sufficient to use m = 3 correction planes for mea-
suring deflections Y1,3 – on the bearing supports, and Y2 – on the gap seal. Additionally,
it is sufficient to chose n = 2 corrections planes for identifying the following imbal-
ances: D1 – on the wheel, and D2 – on the impeller.
As a consequence of numerical simulation using the authors’ software “Forced
oscillations of the rotor”, the results are obtained and tabulated (Table 1).
Using ANN is realized by “Virtual Gene Developer” for the parameters:
(1) topology settings: number of input variables – 9; number of output variables – 2;
number of hidden layers – 2; node number of 2nd (hidden) layer – 18; node
number of 3rd (hidden) layer – 6;
(2) training set: learning rate – 0.01; momentum coefficient – 0.1; transfer function –
hyperbolic tangent; maximum number of training cycle – 5105; target error –
110−5; initialization method of threshold – random; initialization method of
weight factor – random; analysis update interval – 500 cycles.
The related ANN architecture is presented in Fig. 3. Using the abovementioned
software requires transforming all the parameters in dimensionless form with values in
a range [–1, 1]. It can be realized by the following transformations:
j ¼ Dj Yi
D ; Yi ¼ ; ð6Þ
maxfDj g
ISO maxfYj ; Yj g
where DISO
j – the system of admissible residual imbalances according the ISO 1940-1
“Mechanical vibration – Balance quality requirements for rotors in a constant (rigid)
j } = 3.0 kg mm. Additionally, max{Yj, –Yj} =
state”. In the stated case, max{DISO
275.7 lm (Table 1).
Fig. 3. The ANN architecture with related map analysis.

Table 2. The results of learning process and predicted output in dimensionless values
Input data Output Predicted
x1 = 1100 x2 = 1963 x3 = 2215 data output
Y1 Y2
Y3 Y1 Y2 Y3 Y1 Y2 Y3 1
D 2
D 1
D 2
D
0.048 0.004 0.0046 0.233 0.032 0.022 0.371 0.061 0.034 0.33 0.00 0.33 0.00
0.095 0.008 0.009 0.440 0.057 0.040 0.670 0.102 0.059 0.67 0.00 0.67 0.00
0.141 0.011 0.013 0.629 0.078 0.056 0.929 0.133 0.079 1.00 0.00 0.10 0.00
0.004 0.017 0.003 0.035 0.072 0.010 0.072 0.107 0.014 0.00 0.33 0.00 0.33
0.008 0.034 0.005 0.068 0.142 0.019 0.156 0.210 0.023 0.00 0.67 0.00 0.67
0.012 0.061 0.008 0.100 0.211 0.025 0.192 0.309 0.031 0.00 1.00 0.00 1.00
0.052 0.021 0.002 0.260 0.102 0.010 0.417 0.161 0.017 0.33 0.33 0.33 0.33
0.102 0.042 0.004 0.482 0.193 0.017 0.734 0.294 0.026 0.67 0.67 0.67 0.67
0.151 0.062 0.005 0.681 0.279 0.023 1.000 0.412 0.031 1.00 1.00 0.10 1.00
The results of learning process and predicted output in dimensionless values are
presented in Table 2. As a result of learning process and regression analysis, the
following parameters are obtained: sum of error – 2.410−5; average error per output
per dataset – 1.310−6; regression coefficient – 0.999996; slope – 0.998; interception –
3.710−4. Finally, using the regression procedure (4) allows evaluating the dimen-
sionless values of residual imbalances D 1 = 0.361, and D
2 = 0.235.
4 Discussion
The precise system of imbalances includes the following real values: D1 = 1.10 kg
mm, and D2 = 0.70 kg mm. Due to this fact, the common approach for the initial
imbalance DD = 0.10 kg mm allows obtaining the following imbalances: D1 = 1.27
kg mm, and D2 = 0.78 kg mm. This data is insufficient due to the inappropriate rel-
ative errors 15% and 11% correspondently.
However, the results obtained using the proposed ANN method give the following
residual imbalances: D1 = 1.08 kg mm, and D2 = 0.71 kg mm. In this case, the rela-
tive errors are 1.6% and 0.9% correspondently. Consequently, on the example of the
turbopump unit, the effectiveness of using ANN method for ensuring vibration relia-
bility of rotors is completely proved.
Additionally, it can be noted that the further research activities should be aimed at
developing the complemented method and related numerical approach for realizing the
procedure of identification parameters based on the abovementioned analytical
dependencies and method of nonlinear identification of parameters for the compre-
hensive mechanical system “rotor – bearing supports”.
5 Conclusions
As a result, the scientific approach of using artificial neural networks for solving a
significant problem of vibration reliability of rotary machines is proposed, and the
related methodology is specified. This approach includes analytical dependencies,
recent techniques for providing numerical simulations, as well as application of ANN.
The proposed methodology allows refining parameters of the nonlinear mathe-
matical models describing forced oscillations of the rotor as a complicated mechanical
system with nonlinearities. Finally, this approach has a significant advantage in com-
parison with the traditional regression analysis due to the absence of necessity for
setting trial imbalances and carrying out additional initial starts.
Finally, the presented material is proved by the research of rotor dynamics on the
example of turbopump units for liquid rocket engines, as well as the parameters for
ensuring vibration reliability are obtained.
Acknowledgments. The research work reported here was made possible by the research project
No. 0117U003931 ordered by the Ministry of Education and Science Ukraine as a result of joint
cooperation between the Department of General Mechanics and Machine Dynamics, Department
of Manufacturing Engineering, Machines and Tools, and Department of Processes and Equip-
ment of Chemical and Petroleum-Refineries.
References
1. Jardine, A.K.S., Lin, D., Banjevic, D.: Review on machinery diagnostics and prognostics
implementing condition based maintenance. Mech. Syst. Sig. Process. 20(7), 1483–1510
(2006)
2. Gao, Q., Duan, C., Fan, H., Meng, Q.: Rotating machine fault diagnosis using empirical
mode decomposition. Mech. Syst. Sig. Process. 22(5), 1072–1081 (2008). https://doi.org/10.
1016/j.ymssp.2007.10.003
3. Pavlenko, I.V., Simonovskiy, V.I., Demianenko, M.M.: Dynamic analysis of centrifugal
machine rotors supported on ball bearings by combined using 3D and beam finite element
models. In: IOP Conference Series, Materials Science and Engineering, vol. 233, p. 012053
(2017). https://doi.org/10.1088/1757-899x/233/1/012053
4. Pavlenko, I.: Static and dynamic analysis of the closing rotor balancing device of the
multistage centrifugal pump. Appl. Mech. Mater. Trans. Tech. Publ. 630, 248–254 (2014).
https://doi.org/10.4028/www.scientific.net/AMM.630.248
5. Pavlenko, I., Trojanowska, J., Gusak, O., Ivanov, V., Pitel, J., Pavlenko, V.: Estimation of
the reliability of automatic axial-balancing devices for multistage centrifugal pumps.
Periodica Polytech. Mech. Eng. 63(1), 277–281 (2019). https://doi.org/10.3311/ppme.12801
6. Obukhov, O., Smirnov, A., Gusak, O.: Numerical and experimental investigation of the
efficiency of vaned diffuser of centrifugal compressor. In: Institution of Mechanical
Engineers - 8th International Conference on Compressors and Their Systems, pp. 649–658
(2013). https://doi.org/10.1533/9781782421702.11.649
7. Gusak, A., Demchenko, O., Kaplun, I.: Application of small-sized low speed axial stages in
well pumps for water supply. Procedia Eng. 39, 35–42 (2012). https://doi.org/10.1016/j.
proeng.2012.07.005
8. Tarelnyk, V., Konoplianchenko, I., Martsynkovskyy, V., Zhukov, A., Kurp, P.: Comparative
tribological tests for face impulse seals sliding surfaces formed by various methods. In:
Advances in Design, Simulation and Manufacturing. LNME, pp. 382–391. Springer, Cham.
https://doi.org/10.1007/978-3-319-93587-4_40
9. Trojanowska, J., Kolinski, A., Galusik, D., et al.: A methodology of improvement of
manufacturing productivity through increasing operational efficiency of the production
process. In: Hamrol, A., Ciszak, O., Legutko, S., Jurczyk, M. (eds.) Advances in
(2018). https://doi.org/10.1007/978-3-319-68619-6_3
10. Pitel, J., Mizak, J.: Computational intelligence and low cost sensors in biomass combustion
process. In: Proceedings of the 2013 IEEE Symposium on Computational Intelligence in
Control and Automation, IEEE Symposium Series on Computational Intelligence, pp. 181–
184 (2013). https://doi.org/10.1109/cica.2013.6611681. 6611681
11. Antoni, J., Randall, R.B.: Differential diagnosis of gear and bearing faults. ASME J. Vib.
Acoust. 124, 165–171 (2002)
12. Chen, D., Wang, W.J.: Classification of wavelet map patterns using multi-layer neural
networks for gear fault detection. Mech. Syst. Sig. Process. 16(4), 695–704 (2002)
13. Raheem, K.F., Karem, W.: Rolling bearing fault diagnostics using artificial neural networks
based on Laplace wavelet analysis. Int. J. Eng. Sci. Technol. 2(6), 278–290 (2010)
14. Jack, L., Nandi, A.: Fault detection using support vector machines and artificial neural
networks, augmented by genetic algorithms. Mech. Syst. Sig. Process. 16(2–3), 373–390
(2002). https://doi.org/10.1006/mssp.2001.1454
15. Wei, Z., Gaoliang, P., Chuanhao, L.: Bearings fault diagnosis based on convolutional neural
networks with 2-D representation of vibration signals as input. In: MATEC Web of
Conference, vol. 95, p. 13001 (2017). https://doi.org/10.1051/matecconf/201795
16. Jayaswal, P., Verma, S.N., Wadhwani, A.K.: Development of EBP-Artificial neural network
expert system for rolling element bearing fault diagnosis. J. Vibr. Control 17(8), 1131–1148
(2001)
17. Pavlenko, I., Trojanowska, J., Ivanov, V., Liaposhchenko, O.: Scientific and methodological
approach for the identification of mathematical models of mechanical systems by using
artificial neural networks. In: Machado, J., Soares, F., Veiga, G. (eds.) Innovation,
Engineering and Entrepreneurship. HELIX 2018. Lecture Notes in Electrical Engineering,
vol. 505, pp. 299–306. Springer, Heidelberg (2019). https://doi.org/10.1007/978-3-319-
91334-6_41
18. Tian, Z.: An artificial neural network method for remaining useful life prediction of
equipment subject to condition monitoring. J. Intell. Manuf. 23(2), 227–237 (2012). https://
doi.org/10.1007/s10845-009-0356-9
19. Dong, M., He, D., Banerjee, P., Keller, J.: Equipment health diagnosis and prognosis using
hidden semi-Markov models. Int. J. Adv. Manuf. Technol. 30(7–8), 738–749 (2006)
20. Li, B., Yuen, M., James, C.H.: Neural network based motor rolling bearing fault diagnosis.
IEEE Trans. Industr. Electron. 47(5), 1060–1069 (2000)
21. Wang, C.C., Kang, Y., Shen, P.C., Chang, Y.P., Chung, Y.L.: Applications of fault
diagnosis in rotating machinery by using time series analysis with neural network. Expert
Syst. Appl. 37, 1696–1702 (2010)
22. Reddy, M.C.S., Sekhar, A.S.: Application of artificial neural networks for identification of
unbalance and looseness in rotor bearing systems. Int. J. Appl. Sci. Eng. 11(1), 69–84 (2013)
23. Wang, P., Vachtsevanos, G.: Fault diagnostics using dynamic wavelet neural networks.
Artif. Intell. Eng. Des. Anal. Manuf. 15, 349–365 (2001)
24. Pavlenko, I., Simonovskiy, V., Ivanov, V., et al.: Application of artificial neural network for
identification of bearing stiffness characteristics in rotor dynamics analysis. In: Ivanov, V.,
et al. (eds.) Advances in Design, Simulation and Manufacturing. DSMIE-2018. Lecture
Notes in Mechanical Engineering, pp. 325–335. Springer, Cham (2019). https://doi.org/10.
1007/978-3-319-93587-4_34
25. Rojek, I., Kujawińska, A., Hamrol, A., Rogalewicz, M.: Artificial neural networks as a
means for making process control charts user friendly. In: Burduk, A., Mazurkiewicz, D.
(eds.) Intelligent Systems in Production Engineering and Maintenance – ISPEM 2017.
Advances in Intelligent Systems and Computing, vol. 637, pp. 168–178. Springer,
Heidelberg (2017)
26. Yashchenko, A.S., Rudenko, A.A., Simonovskiy, V.I., Kozlov, O.M.: Effect of bearing
housings on centrifugal pump rotor dynamics. In: IOP Conference Series: Materials Science
and Engineering, vol. 233, no. 1, p. 012054 (2017). https://doi.org/10.1088/1757-899x/233/
1/012054
Interaction in Project Management Approach
Within Industry 4.0
Mehmet Cakmakci(&)
Engineering Faculty Industrial Engineering Department, Dokuz Eylül University,

Buca, 35160 Izmir, Turkey
mehmet.cakmakci@deu.edu.tr
Abstract. With the fourth industrial revolution emerging as a requirement of

the global production network, companies and countries are connected more
tightly through supply chain networks around the world. The fourth industrial
revolution “Industry 4.0” means that all innovations in digital technology
mature, especially in the manufacturing sector, during the transformation pro-
cess. These technologies are especially advanced internet, artificial intelligence
and robotics, advanced sensor technology, cloud technology that accelerates
data transfer and utilization, digital fabrication (3D printing), new services and
marketing models, smart phones, and algorithms for navigation tools, delivery
and ride services and autonomous vehicles. All these elements are included in
the global value chain so that they can be shared by many companies from many
countries. So, in the fourth industrial revolution, there are different technologies
that are brought together for different but the same purpose. It is therefore
becoming more complex to organize global organizations that are global,
interconnected, and international. Industry 4.0 brings new dimensions both
qualitative and quantitative within the companies that have used the new
automation technologies for years and have experience and knowledge. Industry
4.0 has a complex nature beyond the modern production management system
(MES) and world class production (WCM). The aim of this study is to show that
the concept of project management in the Industry 4.0 approach, where large-
scale physical and virtual worlds are tried to be brought together, is particularly
affected by this change in the manufacturing sector. With this study, it is stated
that changing the project management approach is a necessity. Moreover, this
interaction is also discussed for industrial facilities.
Keywords: Industry 4.0 Project management

Cyber-physical systems system Project management-industry 4.0 relationship
1 Introduction
The term Industry 4.0 emerged for the first time from a Working Group of the Research
Union BMBF (The Federal Ministry of Education and Research abbreviated BMBF) to
develop the vision of a future industrial society under the influence of the internet [1].
Technological advances have been, since the beginning of the industrial revolution, it
was possible to fold three main strata indicating a large increase in productivity. In the
late 18th century, steam powered machines were used in factories. At the beginning of

https://doi.org/10.1007/978-3-030-18715-6_15
Interaction in Project Management Approach Within Industry 4.0 177
the 20th century, mass production with electric energy was possible, automation has
become widespread in the industry with electronic and information technology [2].
All processes from the start of production to the identification of customer requests
and delivery of the final product are connected to each other faster by cyber-physical
systems, value chains and dynamic data processing technologies in enterprises with
Industry 4.0 (Fig. 1). These trends, sensors, production tools and information tech-
nology industry that goes beyond a single company by preparing the ground for the
systems it value chains. These new correlated systems, called cyber-physical, can
interact with each other using standard internet-based protocols and to define param-
eters and to adapt to changing conditions are analyzing the data. In industry 4.0, these
systems have become widespread, flexible and efficient processes, and higher quality
goods, they will be able to produce at lower cost. The structural changes mentioned,
productivity will increase in production while growth will accelerate in the industry the
workforce profiles will change [2, 3].
According to Geissbauer, the concept of Industry 4.0 is emphasized as the fourth
industrial revolution by “Platform Industry 4.0 in Germany”. In order to adapt to
changing market expectations, the life cycle of products needs to create a new level of
control and organization over the entire value chain, with increasingly individualized,
more specific customer needs. This includes the services that occur during the use of
the customer’s product. The fact that all information about all interconnected processes
in the value chain can be used in real time forms the basis of the fourth industrial
revolution. According to this approach, the ability to obtain the optimal value added
stream at any given time is also vital. The link between all sources and systems,
including people in production, creates dynamic, self-organizing, real-time optimized
value-added connections within and across companies. All of this can be optimized
according to different criteria such as costs, availability and consumption of resources
in the Industry 4.0 implementation [4].
Intelligent production is regarded as a new production model based on intelligent
science and technology that greatly improves the design, production, management and
integration of a product’s entire life cycle [5]. With the use of sophisticated information
and production technologies, products and production processes are optimized from
customer order receipt to delivery of the finished product [6]. It is developed using
advanced materials, intelligent devices and data analysis, adaptive decision making
models and various intelligent sensors in the whole product life cycle to increase
production efficiency, product quality and service level [7, 8]. These technologies are
especially advanced internet, artificial intelligence and robotics, advanced sensor
technology, cloud technology that accelerates data transfer and utilization, digital
fabrication (3D printing), new services and marketing models, smart phones, and
algorithms for navigation tools, delivery and ride services and autonomous vehicles.
The idea of combining automation technology with Lean Manufacturing first
emerged in the mid-1990s with the lean automation of the computer integrated man-
ufacturing (CIM) approach [9]. This leads to the development of a concept that deviates
from
– the big data available in the company,
– the systems already in use,
– the operational requirements,
178 M. Cakmakci
– the different sequencers,

– the organizational structure of the company
– the data available from suppliers,
– the organizational structure of cyber-physical systems
Industry 4.0 has a complex nature beyond the modern production management
system (MES) and world class production (WCM). In this context, the concept of
project management in the Industry 4.0 approach, where large-scale physical and
virtual worlds are tried to be brought together, is also influenced and changed by this
changing. Industry 4.0 technology emerged with the integration of today’s technology
of cyber-physical systems into production planning and control systems, which is the
core of CIM technology.
In intelligent factories that have high automation and flexible production capability,
it is inevitable to judge and make quick decisions of data with embedded software with
large algorithms, where sensors and other sensing systems or devices and large data
gathering are used in the internet of things (IOT), using learning machines, project
managers need to adapt to technological evolution. The aim of this study is to establish
a relationship between Industry 4.0 and project management within the framework of
experience and knowledge on project management and literature review, and to make
proposals so that they can be carried out in terms of quality, cost and time, from the
discovery to the finalization of the project. Industry 4.0 should be understood as the
fourth industrial revolution that will make businesses more productive and competitive,
allowing companies to be more productive and faster to respond to the global market.
The aim of this study is to show that the concept of project management in the
Industry 4.0 approach, where large-scale physical and virtual worlds are tried to be
brought together, is particularly affected by this change in the manufacturing sector.
With this study, it is stated that changing the project management approach is a
necessity. Moreover, this interaction is also discussed for industrial facilities.
2 Methodology
The aim of this scientific study is to analyze and make recommendations to the change
and interaction of the project management philosophy with the introduction of Industry
4.0 in the manufacturing sector, especially in the machinery manufacturing, electronics
sector and automotive sector. For this, a comprehensive literature study was conducted
on reputable databases on both Industry 4.0 and project management. This literature
study was supported on the basis of the knowledge, experience and experience gained
from the application as can be understood from the figures (see Figs. 4 and 5).
Industry 4.0 was firstly dealt with terminologically. The integration of Industry 4.0
into production has been mentioned and in this context, the components of Industry 4.0
have been emphasized. Then the concept of project management was introduced.
Based on the assumption that data and information flow management forms the basis of
project management approach, the information flow in production is discussed and then
project management and Industry 4.0 are associated. Thus, as mentioned above, the
interaction of project management approach with application of Industry 4.0 could be
seen more clearly.
3 Integration of Industry 4.0 in the Production
There are three important features for the integration of Industry 4.0 in production, in
the process from the order of the customer to the delivery of the final product, which
includes the suppliers and all the production components. They are horizontal inte-
gration, end-to-end engineering, and vertical integration.
Fig. 1. The interface between horizontal and vertical integration lies in planning, purchasing,
production and logistics [12]
In the industry 4.0 approach, cross-company and intra-company intelligent cross-

linking and digitization of value creation modules becomes important in the value chain
of the integrated product life cycle and in the value chain of the product life cycle. This
is possible through horizontal integration throughout the entire value creation network
[10, 11]. Through end-to-end engineering throughout the entire product lifecycle,
digitalization and intelligent cross-linking are taking place throughout all phases of the
product life cycle. This process encompasses all processes from raw material acqui-
sition to the manufacturing system, from product use to the end of the life of the
product.
In the manufacturing process where engineering, design, manufacturing, marketing,
and sales activities are interrelated, vertical integration provides seamless connectivity
among all elements included in an organization’s product life cycle. Digitalization
within the different aggregation and the intelligent cross-linking and hierarchical levels
of a value creation module from manufacturing stations via manufacturing cells, lines
and factories can be integrating by the vertical integration (see Fig. 1).
4 Supporting Technologies Industry 4.0 in the Production
With Industry 4.0, products, manufacturing, technology, workplace and people will be
more integrated with each other. With integrated manufacturing, machines and devices
are no longer isolated from the system, but are treated as parts of the system, so that all
components can be coordinated effectively to obtain advanced components. We can
also incorporate the society in the national and international market dimension into this
integration [13]. In this context, in the structure of the Industry 4.0, six important
180 M. Cakmakci
features to be considered in terms of project management come to the forefront. They

are internet of things (IOT), cyber-physical systems (CPS), big data, automation,
machine learning and cloud computing.
4.1 Internet of Things (IOT)

Today, many physical objects used in production can be connected via embedded
software, sensors, electronic and network devices, as well as through Internet tech-
nology developments. In this context, for integrated and intelligent manufacturing
systems, Internet of Things offers a unique and highly needed foundation that can
connect all elements of a production system together [13]. The efficiency of collecting
data with the Internet of Things is increased and at the same time the quality of the data
is significantly improved. Internet of Things (IOT) also provides network control and
management of production equipment and assets and information flow, as well as new
advanced services based on interactions between robots and objects.
4.2 Cyber-Physical Systems

The industry 4.0 approach requires knowledge technology and software technology to
be combined with and coordinated with mechanical components in the customizer
MES production. In this way, data transmission and exchange via the Internet, which
constitute the infrastructure, takes place in real time. The basic components are mobile
and mobile devices, devices and machines (including robots), embedded systems and
network-connected objects (Internet of Things). This is why cyber physical systems
have a central function in the Industry 4.0 approach [14, 15]. In this context, these
systems, which are used integrated, are called cyber physical systems. Cyber-physical
systems, which we can define as physical and engineering systems, are systems of
computational assets that are intimately linked to the surrounding physical world and
its ongoing processes, while at the same time offering and cooperating with data access
and data processing services [16, 17].
4.3 Big Data

In the customizer MES production process where cyber-physical systems (CPS) and
internet of things (IOT) systems are used, massive amounts of data arise as a result of
machines, robots, logistics and user feedback. For the purpose of the industry 4.0
approach, it has become very important to evaluate these data and to make real-time
deductions from the data. When evaluating the data, approaches such as machine
learning need to be used in terms of both size and structure. With these characteristics
of production systems, the control of production systems will be accelerated and
proactive decisions will be made in the process management [18]. In this context, the
characteristics of project managers within the scope of project management at all stages
of production will be increased in accordance with these developments.
4.4 Automation
Automation, robots and new production technology are the unavoidable elements of
future production. During the transition period and shortly before of the industry 4.0
implementation, automation and utilization of robots such as transportation, line
feeding and material handling within a facility were and will be of relevance in different
areas of the production processes. Stock et al. [11] pointed out automation can be
considered one of the main trends and expected developments within the Industry 4.0
concept. In addition to these technologies within Industry 4.0, with the use of additive
production technologies such as 3D printing, cloud technology, production processes
are accelerated, quality is improved and performance is increased [19]. In this regard,
automation is one of the most important features for project management and especially
project planning in Industry 4.0 framework.
4.5 Machine Learning

In order to be more effective in the interaction between Industry 4.0 and industrial
internet, communication between all machines, robots, equipment and devices involved
in production must be of high quality and problems must be minimal. In addition, it is
desirable to make inferences from previous actions in order to take the necessary steps
in the future in order to solve the problems between machines, equipment and devices
automatically. In other words, all machinery, robots, equipment and devices involved
in production are required to learn from each other in relation to each other. We can
express machine learning ability to understand and learn the content of a physical
system through computational algorithms based on computer based on learning ability,
which is one of the basic characteristics of human intelligence. Applying the machine
learning algorithm, which requires the use of algorithms that allow machines to sim-
ulate human intelligence, allows a machine or a robot or an equipment or a device to
automatically learn its own fundamentals and operating conditions. Thanks to advances
in sensor fusion, machine learning, edge and cloud computing, sees adoption in many
ways and now plays a central role in intelligent factory predictive analysis [20].
4.6 Cloud Computing

Cloud computing provides an Internet-based computing service that makes it possible
for a user to share software without having to install the required software locally. In
the near future, with the progress of intelligent manufacturing technology, it can be
predicted that data and information will be collected in real time by well-equipped
sensors and converters from all fields in the product life cycle. These collected data can
be processed with the help of cloud computing technology. Then will be accurate
decisions can be made continuously and autonomously with little or no human
intervention [21].
182 M. Cakmakci
5 Seamless Information Flow
In the context of decisions making and evaluating real-time data gathered in the main
sub-processes of a project such as application, monitoring and control of the project,
the internet, advanced computers, embedded software, cloud computing technology,
well-equipped sensors and converters are at the forefront.
Future plants will be digitized with a high level of integration among the various
subsystems and characterized by an uninterrupted flow of information using real-time
planning and control [16, 22]. A highly digitized and vertically integrated factory will
allow decision-making based on real-time information, improving production planning
and control activities. Real-time information access provides continuous, real-time
scheduling and control of production operations [23]. Passing on real-time controls
requires new conceptual models for planning and control. Real-time control is
implemented today at the machine and production line level. However, at planning
levels, including planning in MES systems, the existing concepts are based on cyclic
data processing and rescheduling.
Industry 4.0 technologies have the potential to provide real-time planning and
control of all planning activities. Vertical integration is a key feature of Industry 4.0
[13], providing a holistic and integrated knowledge management that can improve
production logistics, from workshop to horizontal integration, to different subsystems
and enterprise resource planning (ERP) systems. The integration of IT systems and the
digitalization of production in the context of production logistics, the tasks required for
the production planning and control (PPC) and the routing of the material flow at the
plant are carried out with the support of IT systems. This requires the implementation
of the necessary systems first, then the use of the systems.
The integration of IT systems is necessary to fully ensure the potential benefits of
Auto ID technology. Real-time control of production through an radio-frequency
identification (RFID) enabled workshop, an MES system or an ERP system, requires
that information obtained from the identification of objects be communicated to the
higher-level IT system.
Cloud computing is essentially “on-demand” IT services. This addition of pro-
duction makes service-based production transition known as cloud production possible
[24–27]. Companies’ production resources and capabilities can be associated with
potential customers of production services through cloud computing. It can analyze
system requirements and propose a service package that extends from product design,
production, testing and other production capabilities to the product life cycle. This link
between all the production resources in a network and the customer’s specific
requirements gives the ability to use all resources better to deliver the desired output.
Both external and internal logistics can be optimized according to requirements for
production resources and capabilities [28].
6 Project Management and Industry 4.0 Relationship

Concept
Project Management is the discipline of organizing and managing resources in

accordance with the client’s expectation to complete a project in terms of quality, time
and cost constraints. As defined by the Project Management Institute (PMI), a project is
a temporary effort to create a unique product, service or outcome [10]. This charac-
teristic of being a transitional attempt is incompatible with processes or processes
which are continuous or semi-continuous functional work to repeatedly create the same
product or service.
Project management has been implemented for thousands of years. Resources from
a variety of skill areas must be drawn together towards a common goal. Targets are
carefully drawn, including performance criteria, schedules and budgets.
During the period from World War II to the present day, project management was
used to manage all the big and complex projects in all the sectors, especially in
construction, machinery, chemistry, shipbuilding and aviation. With the use of project
management approach, network methods, gantt chart diagram, critical path method
(CPM), program evaluation and review technique (PERT), acceleration of projects
method, earned value analysis (EVA), work breakdown structure (WBS) are used.
With the larger and more complex features of the projects, simulation and artificial
intelligence techniques have been used in project management together with advanced
computer use. Project management requires extensive planning and coordination.
A Project Manager is responsible for managing the resources of large projects [29].
Intelligent manufacturing (also known as smart manufacturing) is a broad concept
of manufacturing with the purpose of optimizing production and product transactions
by making full use of advanced information and manufacturing technologies (see
Fig. 2) [5, 30–32]. An MES (Manufacturing Execution System) is an IT system that
can integrate production, material, quality, process and maintenance data in a single
platform. In real-time, the data is sent to the entities and receives data from them. Being
able to manage integrated data in a single database allows the information to be
transformed into an entry that is not stored passively, but causes more.
Fig. 2. Industry 4.0 concept [16, 32]

184 M. Cakmakci
In the Industry 4.0 approach, businesses are increasingly dedicated to regional and
global virtual value chains, and all business activities and processes are supported by
well-designed and implemented digital communication technologies and systems [34].
Today, internet is used as the most important communication tool and information
source. Virtual representations of the real world are being produced and used, and
cyber systems are becoming more common, so that the business world is becoming
more and more digital. In this context, Industry 4.0 describes a development that is
expected to lead to a new fourth industrial revolution and is characterized mainly by
digitalization in the sectors of mass production in particular [22]. As a reflection of this,
it is observed that almost all companies are affected by digitalization and increasingly
tend to realize their benefits fully technologically, economically and managerially (see
Fig. 3).
Fig. 3. Machines communicate man-machine [33]
Under Industry 4.0, production systems are updated to an intelligent level. Intel-
ligent production is used to support intelligent production that utilizes advanced
information and production technologies to achieve flexible, intelligent and reconfig-
urable manufacturing processes that include core technologies such as IOT, CPSs,
cloud computing, large data analysis (BDA) and information and communications
technology (ICT) [1].
Intelligent manufacturing is a broad production concept for optimizing production
and product operations using fully developed information and production technologies
[33]. A typical product is considered to be a new production model based on intelligent
science and technology that has greatly improved the design, production, management
and integration of the entire life cycle. The entire product lifecycle can be facilitated
using a variety of intelligent sensors, adaptive decision making models, advanced
materials, intelligent devices and data analysis [7, 8]. Thus, production efficiency,
product quality and service level are developed [35]. In this context, the human factor
and human machine interaction (HMI) are considered as one of the main beneficiaries
in all these visions (see Fig. 4) [36].
Since the mid 1980s, when CIM-Computer Integrated Manufacturing technology
and Industry 4.0 concept have started to be used as ideas, production processes and
procedures in modern companies including global production network have been
developed and changed in this period. With digitization, both traditional manufacturing
companies and service sector businesses have created more simplified data manage-
ment, less costly and more personalized solutions, greater automation of labor intensive
processes, or greater opportunities to take measures that simplify these processes. With
Industry 4.0, companies already involved in the digitization process and who need to be
able to manage this process have become more prominent in project management.
Project managers affected by digitization have continued to organize their internal
structures using digital products in order to adapt to this change.
Fig. 4. Interaction environment between Industry 4.0 and Industrial Internet: visualization,
visual analytics, and human-machine interaction [36]
7 Evolution of Project Management Approach Within

the Scope of Industry 4.0
Until the end of the twentieth century, the classical project management approach has
been in the construction industry (industrial facilities, roads, bridges, dams, thermal
power plants, hospitals, large residential projects, planes, airplanes, trains), all these
processes in the manufacturing sector can be seen clearly. In other words, participation
186 M. Cakmakci
and contribution of the human factor to the actions of the project are clearly seen and
observed in all processes of the project management.
Project actions should be checked from project announcement meetings to project
feasibility analysis, project action and tasks are defined, timed, and used as an
instrument in project management and control, depending on the planning of the
project in coordination with the project planning and finally includes all activities up to
the termination of the project. The initiation, planning, implementation, monitoring and
control of the project, as well as project closure processes, constitute the sub-processes
of the classical project management approach.
Industry 4.0 represents a new organization and control path for full value added
systems. The main goal is to meet individual customer needs with mass production
costs. For this reason, order management affects all areas such as research and
development, production, commissioning, delivery and recycling of manufactured
products. The new opportunities are digitization of production with the help of the
base, cyber-physical production systems. For this reason, resources such as all
employees, products, resources and systems must be integrated as intelligent, self-
organizing, inter-company, real-time and autonomously optimized examples [16, 23].
Nowadays, in the manufacturing sector, which is the pioneering sector in which the
Industry 4.0 approach is applied, the human factor is gradually being replaced by
Fig. 5. Project management process groups [10, 31]

robots and full automations in the sub-processes of applying, monitoring and con-
trolling the processes, especially in the automotive and electronics sectors (see Fig. 5).
The functions of these processes are transferred to robots and machines. In this case,
the initiation and planning processes gain more importance than the functions of the
classical project management approach. That is, the tasks to be taken from the robot
during the implementation, monitoring and control of the project and the tasks given to
the robot and the machine must be designed in accordance with the digitization pro-
cedures of the job descriptions, taking into account the processes from taking the order
from the customer to delivering the final product to the customer. In other words, the
identification of the human factor involved in the initiation and planning processes of
the project needs to be restructured to include its functions during the implementation,
monitoring and control of the project.
8 Conclusion
In the manufacturing sector, project design and planning in virtual value chains with
national and international qualities created by digital communication technologies will
be performed in a similar manner to classical project management. However, these
processes will certainly change as more intelligent learning systems become more
involved in the production process over time. In these multi-participating virtual value
chains created, information and information flow will become more intense and project
management becomes more difficult due to decision-making. At the strategic, tactical
and operational level, further initiatives will need to be shared in order to decide on the
process from the selection of the customer from the market selection to the selection of
the material supplier, the selection of the production line, and the determination of the
quality control methods to be used. While a central decision-making approach is
employed in classical project management, a decentralized project management
approach is adopted in Industry 4.0 [37, 38].
In this context, although all the stakeholders involved in the virtual value chain of
the project have hierarchical authority, their management will also differ from the
classical project management. Project team members and managers will need to have
facilities that can use digital communication technologies, cyber physical systems. It
should be discussed how this interaction in industrial plant, bridges, dams, nuclear and
thermal power plants, hospitals, shopping centers, aviation sector, shipbuilding and
large residential projects can be discussed while the interaction between project team
members and managers’ developments in the manufacturing sector is like this.
References
1. Sheer, A.-W.: Industrie 4.0: Von der Vision zur Implementierung, White paper Nr. 5,
Saarbrücken, Germany (2015)
2. Sanders, A., Elangeswaran, C., Wulfsberg, J.: Industry 4.0 implies lean manufacturing:
research activities in industry 4.0 function as enablers for lean manufacturing. J. Ind. Eng.
Manag. 9(3), 811–833 (2016)
188 M. Cakmakci
3. Tansan, B., Gokbulut, A., Targotay, C., Eren, T.: Turkiye’nin Kuresel Rek abetciligi icin Bir
Gereklilik Olarak Sanayi 4.0. TUSIAD - BCG the Boston Consulting Group, Tusiad-
T/2016-03/576 (2016)
4. Geissbauer, R., Schrauf, S., Koch, V., Kuge, S.: Industry 4.0 – Opportunities and Challanges
of the Industrial Internet. Pricewaterhouse Coopers AG Wirtschaftsprüfungsgesellschaft,
Germany (2014)
5. Zhong, Y.R., Klotz, E., Xu, X.: Intelligent manufacturing in the context of industry 4.0: a
review. Engineering 3(5), 616–630 (2017)
6. Kusiak, A.: Intelligent Manufacturing Systems. Prentice Hall Press, Old Tappan (1990)
7. Li, B., Hou, B., Yu, W., Lu, X., Yang, C.: Applications of artificial intelligence in intelligent
manufacturing: a review. Front. Inf. Technol. Electron. Eng. 18(1), 86–96 (2017)
8. Davis, J., Edgar, T., Porter, J., Bernaden, J., Sarli, M.: Smart manufacturing, manufacturing
intelligence and demand-dynamic performance. Comput. Chem. Eng. 47, 145–156 (2012)
9. Kolberg, D., Zühlke, D.: Lean automation enabled by industry 4.0 technologies. IFAC-
PapersOnLine 48(3), 1870–1875 (2015)
10. Project Management Institute: A Guide to the Project Management Body of Knowledge, 5th
edn. Project Management Institute Inc., Newtown Township (2013)
11. Stock, T., Seliger, G.: Opportunities of sustainable manufacturing in industry 4.0.
Procedia CIRP 40, 536–541 (2016)
12. Wolter, M.I., Mönnig, A., Hummel, M., Schneemann, C., Weber, E., Zika, G., Helmrich, R.,
Maier, T., Neuber-Pohl, C.: Industry 4.0 and the consequences for labour market and
economy scenario calculations in line with the BIBB-IAB qualifications and occupational
field projections. IAB-Forschungsbericht 8 (2015)
13. ILX Team: 4(.0) Things for Project Mangers to Look out for in the Fourth Industrial
Revolution (2017). https://www.ilxgroup.com/eur/individual/blog/4-things-for-project-mang
ers-to-watch. Accessed 30 Apr 2018
14. Bendel, O.: Cyber-physische systeme, definition. https://wirtschaftslexikon.gabler.de/definit
ion/cyber-physische-systeme-54077. Accessed 30 Apr 2018
15. Sheer, A.-W., Mattheis, P., Steinmann, D.: PPS, CIM Handbuch - Geitner U. W.
(Herausgeber). Friedr. Vieweg & Sohn Verlagsgesellschaft mbH, Braunschweig (1987)
16. Monostori, L., Kádár, B., Bauernhansl, T., Kondoh, S., Kumara, S., Reinhart, G., Sauer, O.,
Schuh, G., Sihn, W., Ueda, K.: Cyber-physical systems in manufacturing. CIRP Ann.-
Manuf. Technol. 65, 621–641 (2016)
17. Lee, E.A.: Cyber physical systems: design challenges. University of California at Berkeley:
Electrical Engineering and Computer Sciences (2008)
manufacturing systems. University of Cincinnati, University Cooperative Research Center
on Intelligent Maintenance Systems (2014)
19. Kagermann, H., Lukas, W., Wahlster, W.: Industrie 4.0 - Mit dem Internet der Dinge auf
dem Weg zur 4. Industriellen Revolution, VDI Nachrichten, 13 (2011)
20. Isaacs, D., Astarola, A., Diaz, J., Arejita, B.: Making factories smarter through machine
learning. IIC J. Innov. (2017)
21. Yubao, C.: Integrated and ıntelligent manufacturing: perspectives and enablers. Engineering
3, 588–595 (2017)
22. Kelly, W.: The future of digital business: what does this mean for project teams? (2017).
https://www.iapm.net/en/news-events/iapm-blog/!/show/266/industry-40-and-project-manag
ement/. Accessed 14 May 2018
23. Bauernhansl, T., Hompel, M., Vogel-Hauser, B.: Industrie 4.0 in Produktion, Automa-
tisierung und Logistik - Anwendung - Technologien – Migration. Springer, Wiesbaden
(2014)
24. Shen, W.M., Norrie, D.H.: Agent-based systems for intelligent manufacturing: a state-of-the-
art survey. Knowl. Inf. Syst. 1(2), 129–156 (1999)
25. Lu, B.H., Bateman, R.J., Cheng, K.: RFID enabled manufacturing: fundamentals,
methodology and applications. Int. J. Agile Syst. Manag. 1(1), 73–92 (2006)
26. Zhong, R.Y., Li, Z., Pang, L.Y., Pan, Y., Qu, T., Huang, G.Q.: RFID-enabled real-time
advanced planning and scheduling shell for production decision making. Int. J. Comput.
Integr. Manuf. 26(7), 649–662 (2013)
27. Huang, G.Q., Zhang, Y.F., Chen, X., Newman, S.T.: RFID-enabled real-time wireless
manufacturing for adaptive assembly planning and control. J. Intell. Manuf. 19(6), 701–713
(2008)
28. Strandhagen, J.W., Alfnes, E., Strandhagen, J.O., Vallandingham, L.R.: The fit of Industry
4.0 applications in manufacturing logistics - a multiple case study (2017). https://brage.
bibsys.no/xmlui/bitstream/handle/11250/2473354/AIM%2BManuscript%2Brevised.pdf?seq
uence=2. Accessed 15 Apr 2018
29. Gentile, S.: BRU MSMPP WP, Project Management Information Technology Industry
(2012)
30. Kagermann, H., Wahlster, W., Helbig, J.: Recommendations for implementing the strategic
initiative INDUSTRIE 4.0: final report of the industrie 4.0 Working Group. Frankfurt/Main,
National Academy of Science and Engineering (2013)
31. Tiede, T.: PMI Project Management Priciples (2014). https://www.slideshare.net/tltiede/pmi-
project-management-principles. Accessed 24 May 2018
32. Bauernhansl, T.: Industry 4.0: challenges and opportunities for the automation industry. In:
7th EFAC Assembly Technology Conference 2013, Davos, Switzerland (2013)
33. Kazutoshi, M.: Industry 4.0: bringing the human-machine relationship to the next level.
Genex Partners, Japan. https://www.twynstragudde.nl/sites/default/files/content/blog/s0984_
cww_genex-genexversion-hr.pdf. Accessed 15 Apr 2018
34. Semolic, B., Steyn, P.: Industry 4.0: virtual value chains and collaborative projects. PM
World J. 6(9), 1–20 (2017)
35. Kagermann, H., Anderl, R., Gausemeier, J., Schuh, G., Wahlster, W. (eds.): Industrie 4.0 in a
Global Context-Strategies for Cooperating with International Partners. http://www.acatech.
de/publikationen. Accessed 04 May 2018
36. Posada, J., Toro, C., Barandiaran, I., Oyarzun, D., Stricker, D., De Amicis, R., Pinto, E.B.,
Eisert, P., Döllner, J., Vallarino Jr., I.: Visual computing as a key enabling technology for
industrie 4.0 and industrial internet. IEEE Comput. Graph. Appl. 35(2), 0272–1716 (2015)
37. Hofmann, E., Rüsch, M.: Industry 4.0 and the current status as well as future prospects on
logistics. Comput. Ind. 89, 23–34 (2017)
38. Smit, J., Kreutzer, S., Möller, C., Carlberg, M.: Policy department: economic and scientific
policy-industry 4.0, industry 4.0 analytical study. European Union (2016)
Applying Scrum in New Product
Development Process
Jan Betta(&), Tomasz Chlebus, Dorota Kuchta,

and Agnieszka Skomra
Faculty of Informatics and Management,

Wroclaw University of Science and Technology, Wrocław, Poland
jan.betta@pwr.edu.pl
Abstract. For all organisations New Product Development (NPD) [1, 2], is a
sophisticated process, necessary for introducing successfully the product on the
market. Successfully means: customer satisfaction, benefits for organisation,
new organisational and personal skills and experiences. The NPD process fulfils
all conditions of project’s definition; that’s why many authors present the
possibilities of applying to NPD process management instruments, specific for
project management. Such attempts are limited to classical project managements
methodologies and tools. Recently we observe growing interest in Scrum
methodology, [3, 4]. Introduced for IT project management, nowadays is
transferring completely or partially to manage the projects in other than IT fields
of activities - healthcare, finance, consulting, education, and others, giving good
results. This paper is a conceptual one, presenting an attempt made to show the
possibility, utility and manner of transferring principal elements of Scrum
toward NPD process. The objective of the paper is turn spotlight of New Product
Development specialists on such possibility and encourage them to apply Scrum
in this process management. The objective is achieved mainly in Sect. 5, where
for all steps and activities in NPD process, their classification has been done,
dividing them on unique or cyclical. The last ones (they are in the majority) can
be managed according to Scrum methodology. It has been shown, for respective
Scrum elements: roles, events and artefacts, how they can work in NPD process
management. This study needs further research and empirical verification.
Keywords: New product development Agile Scrum methodology
1 Introduction
Nowadays we are witnesses of the still rising quantity of new products. The drive
wheel of this phenomenon are still varying costumers’ needs, encouraging the pro-
ducers to put on the market almost constantly new products. New product development
process is commonly recognized as a project, what gives the possibility to use project
management instruments for this process. For example, such classical tools like Work
Breakdown Structure, Scheduling, resources and cost planning or risk management are
commonly used, [5]. From about twenty years classical methodologies of project
management like PMI, IPMA or Prince2 turned out to be insufficient for more and
more sophisticated projects, mainly in IT. As a response, Agile approach appeared; it
https://doi.org/10.1007/978-3-030-18715-6_16
Applying Scrum in New Product Development Process 191
the frame of it there are several methodologies. Mostly used in IT projects is Scrum
methodology. From 3–4 years Scrum has been transferred completely or partially to
manage the projects different from IT, giving satisfying results [6–12]. In this paper the
authors present justification, possibility and advantages of using the basic elements of
Scrum to manage the New Product Development process. The main objective of the
paper is to turn spotlight of New Product Development specialists on such possibility
and encourage them to apply Scrum in this process. In consecutive Sects. 2, 3 and 4 the
lectures will find: description of NPD process, agile approach to project management,
Scrum methodology brief presentation and, finally, in the central Sect. 5 – the proposal
of Scrum elements transfer toward NPD. Finally, Sect. 6 deals with summarising of the
contain and suggestions of main directions to develop the subject. The paper is a
proposal for NPD specialists to try apply the presented approach in real cases.
2 New Product Development Process
One of the key processes Innovation is the New Product Development Process – NPD
[1] which is present below. The New Product Development process is often referred to
the Stage-Gate innovation one, developed as a result of comprehensive research on
reasons why products succeed and why they fail. When the teams collaborate in
developing new innovations, proceeding according to the following eight steps, the
product is marketable, relatively quickly achieved and accurately made.
Step 1: Generating
During the NPD process, keep the system nimble and use flexible discretion over
which activities are executed. One may want to develop multiple versions of your road
map scaled to suit different types and risk levels of projects.
Step 2: Go or no go?
Specific criteria for ideas that should be continued or dropped. Stick to the agreed upon
criteria so poor projects can be sent back to the idea-hopper early on. Because product
development costs are important, it means taking the top 3 competitors’ innovations
into account.
Step 3: Testing the Concept
It is important to note, it is different from test marketing. Apart from patent research,
design due diligence, and other legalities involved with new product development.
Knowing where the marketing messages will work best is often the biggest part of
testing the concept. Does the consumer understand, need, and want the product or
service?
Step 4: Business Analysis
The activity in this step consist in building a system of metrics to monitor progress. It is
important for an organization to observe the criteria and metrics.
Step 5: Marketing Tests
Arranging tests groups, launching versions, and then forming test panels after the
product or products have been tested ere the activities included in this step.
192 J. Betta et al.
Step 6: Technicalities & Product Development

In this step, the production department will make plans to produce the product. The
marketing department will make plans to distribute the product. The finance department
will provide the finance for introducing the new product.
Step 7: Commercialize
At this stage, consumers are purchasing the good or service, and technical support is
monitoring progress. Keeping the distribution pipelines loaded with products is an
integral part of this process. Keeping the product’s name firmly supplanted into the
minds is the second one.
Step 8: Post Launch Review and Perfect Pricing
In this final stage, overall value relevant to cost of goods sold is evaluated, making sure
internal costs aren’t overshadowing new product profits. You continuously differentiate
consumer needs as your products age, forecast profits and improve delivery process.
The entire new product development process is an ever evolving system with
possible errors, designs trashed, and loss. Productivity during product development can
be achieved if, and only if, goals are clearly defined along the way and each process has
the possibilities clearly defined.
There is relatively little in the literature on the important NPD question: how an
organisation can define and implement behaviour, structures and processes necessary to
make good practices of NPD working. Six key design elements which define beha-
vioural conditions necessary for a successful NPD process are proposed [2]. As we will
see in the Sects. 3 and 4, they are encouraging to use Scrum methodology elements [3].
Combining the NPD process with Scrum main elements is the first important and
helpful step to introduce good practices into NPD working. The process of imple-
menting a New Product Development is sophisticated, takes some defined time and is
divided into several phases. Depending on kind and complexity of products, magnitude
of production and organisational as well as technical preparation, different method-
ologies of project management are applied [5].
3 The Agile Approach to Project Management
The description of an agile approach to project management is essentially based on [3]

and the earlier authors’ paper [6]. Classical methods of project management are based
on a stage-based approach to product development (especially software development).
Classical methods assume that the transition to each succeeding stage becomes possible
following completion of work at the previous stage. In this approach, product
requirements are defined at the planning stage and are not subsequently modified. It is
assumed that the conditions governing project team operations are stable. As a result,
there is a strong focus on documenting actions, which becomes the main aspect of
control of project execution [3, 6]. One of the most popular examples of classical
methods of project management is the waterfall model, which consists of seven stages
of software development for the delivery of a software product to the customer [3].
The agile approach to project management, also called the adaptive approach, was
created in 2001 by a group of developers who prepared Manifesto for Agile Software
Development, a credo of agile software development. The Manifesto consists of four
demands [4]:
– ‘individuals and interactions over processes and tools,
– working software (product) over extensive documentation,
– customer collaboration over contract negotiation,
– responding to change over following a plan’.
The word ‘over’, as used in the Manifesto, does not mean that the concepts regarded as
classical have been given up completely. The authors of the Manifesto emphasise that
‘while there is value in the items on the right, more valuable are the items on the left
one’ [3].
The agile method (approach) focuses on being open and reactive to changes in
order to ensure for the client a product that complies with expectations. This approach
is based on iterative development and delivery of the product to the customer. One of
its most important aspects is that the client is a part of the project team and co-operates
with this team throughout project operations. The centre of attention is the execution of
work, whereas documentation is limited to the minimum and replaced with frequent
project meetings, with special emphasis on direct meetings. Frequently these meetings
have an influence on limited control, thanks to the relationships between members of
the project team based on engagement, co-operation, a sense of responsibility for the
work to be performed, and the product, as well as on mutual trust. The core element is a
self-organising team which independently makes all decisions regarding the method of
operation, taking into account necessary adaptations for changing conditions of func-
tioning [6].
The agile approach to project management is currently being implemented in an
increasing number of activities, such as information technology, healthcare, finance,
consulting, education, and others [6–9]. The most popular agile method is Scrum. Over
58% of agile practitioners report that they use this approach in project execution [3].
The research shows also that 54% of respondents use Scrum in combination with other
practices, while 42% emphasise exclusive use of Scrum.
4 Scrum Methodology
The Scrum methodology was formulated in 1995 by Schwaber and Sutherland [3].
Scrum is a framework designed to overcome complex adaptive problems and to deliver
a product with the greatest possible value for the customer. Scrum is based on
empiricism, which builds:
– clarity within each process,
– inspection to detect problems in the project,
– adaptation to changes.
Scrum consists of four main elements: Roles, Events, Artefacts and Rules.
194 J. Betta et al.
4.1 Roles [3]

There are four main roles in Scrum. Scrum Master is the person responsible for the
understanding and use of the values and rules of Scrum by the Development Team and
Product Owner. The main duties of the Scrum Master are to serve the Scrum Team in
order to achieve project aims and to ensure that the values of Scrum are applied
properly by the Scrum Team.
The Product Owner is the person who is familiar with the business associated with
the project and responsible for maximising the value of the product. The main duties of
the Product Owner are to control and manage the Product Backlog [Elements in Scrum,
The Artefacts].
The Development Team is the team responsible for developing the product
according to requirements. The Development Team is ‘self-organising’, which means
that it has a high degree of autonomy.
The Scrum Team comprises the Scrum Master, Product Owner, and Development
Team.
4.2 Events [3]

Events in Scrum are important in order to provide regularity in Scrum. Events in Scrum
are Sprint, Sprint Planning, Daily Scrum, Sprint Review and Sprint Retrospective.
The Sprint is a limited time interval oriented towards an increment in functionality
of the project product. The Sprint usually lasts about 30 days or less and consists of the
following events: Sprint Planning, Daily Scrum, Sprint Review, and Sprint
Retrospective.
The Sprint Planning - the main goal of this event is to establish the scope of work to
be done during the iteration. During Sprint Planning, items from the Product Backlog
are selected; these items will be implemented to ensure the creation of product
increment.
The Daily Scrum is a daily 15-min meetings of Scrum Master and the Development
Team. The Daily Scrum is vital for monitoring progress in Sprint and for detecting
problems that may threaten achievement of the goal of the Sprint.
The Sprint Review is an event at the end of every Sprint to inspect the delivered
functionality (Increment) and implement the Product Backlog.
The Sprint Retrospective consists of reflections on the completed Sprint with some
projection for the next one.
4.3 Artefacts [3]

Artefacts the material or immaterial results of the work, which enable inspection and
adaptation in Scrum methodology.
The Product Backlog determines the scope and sequence of a list of features which
should be implemented during the project. The document is open and changes can be
introduced at any stage of the project.
The Sprint Backlog is a part of the Product Backlog. It is created from Product
Backlog items selected for the Sprint; the Sprint Backlog is created and managed by the
Development Team.
The Increment defines the complete components of the Product Backlog completed
during the Sprint and other previous Sprints.
The Definition of Done focuses on clearly understanding when the element from
Product Backlog can be accepted as finished.
4.4 Rules [3]

Rules are defined as being linked with the method; they define relationships between
Roles, Events and Artefacts.
Figure 1 presents iteration in Scrum methodology. Scrum is characterised by a
specific development process based on incremental growth of the product and iterations
that are fixed in time [3, 6]. The first step is to create a Product Backlog, where all
requirements for the project are kept. During every Sprint, elements from the Product
Backlog which are compatible with the aim of the Sprint are selected for the creation of
the Sprint Backlog. The Sprint is a fixed period of time during which the Development
Team works to provide new functionality for the customer. The Sprint begins with
Sprint Planning. In this Event, the Development Team selects tasks for the Sprint
Backlog and evaluates them. During the Sprint, the Daily Scrum, or daily meeting of
the Scrum Master and Development Team, is held. The total number of finished tasks
from the Product Backlog make up the Increment, which should be usable by the
customer. The Sprint Review and Sprint Retrospective are conducted by the Scrum
Team at the end of each Sprint [3].
Fig. 1. The Sprint [8].

196 J. Betta et al.
5 Transfer of Scrum Elements Toward NPD
As it was shown in the Sect. 3, we observe the increasing number of fields of human
activities and situations where the applications of selected elements of Scrum is pos-
sible and useful. In this section – central in the whole paper, the authors show why and
how the Scrum methodology can be helpful in NPD process.
In the context of NPD project should be understood as a set of ordered activities
planned at lead to obtain the new product, ready to be manufactured and purchased,
accomplishing all requirements and constraints (delay, cost, technical, quality and
others).
Basing on earlier studies [7–9], the authors propose three main roles based on the
Scrum methodology:
– NPD Scrum Master - a “servant-leader”; he or she eliminates impediments on the
way to the goal. He or she coordinates the activities of members of the NPD Team
and:
• support the NPD Team in the realization of activities;
• support meetings;
• support in the understanding of the principles of planning activities in an
empirical environment;
• ensuring free communication between the NPD Team and the NPD Product
Owner.
– NPD Product Owner - a representative of the enterprise Top Management, having
decision-making power in the scope of undertaken actions during the NPD.
– NPD Team - team consisting of specialists responsible for the implementation of all
activities needed in NPD Process. In its composition, there should be people from
all of Enterprise units, involved in the NPD.
In addition, two Artefacts are pointed out [7–9]:
– NPD Product Backlog - includes all activities defined in the NPD Process that are
necessary to be carried out. It includes two types of activities:
• Cyclic activities - activities in the project/NPD Process that are realized con-
tinuously, at specific intervals;
• Unique activities - activities undertaken once during the project.
It is also recommended to assess the labour intensity of tasks and include them in the
NPD Product Backlog.
– NPD Sprint Backlog - activities (cyclic and/or unique) selected from the Product
Backlog to be implemented during the one iteration.
The classification of actions of eight steps described in Sect. 2. for cyclic and unique
activities is proposed below.
Step 1: Generating
This step needs a lot creativity of people working on it, their continuous cooperation
and information exchanges which should lead to the continuous improvement.
Therefore it is recommended that the activity of this step should be a cyclical one (NPD
Sprint, Fig. 1). Intervals between activities should be adapted to the needs of the NPD
Team, NPD Product Owner and the dynamics of changes in the environment. The most
important here is a human factor, which is a part of the postulates of agile management.
Building a team based on mutual trust, cooperation and commitment will enable quick
decision making and adaptation to unusual conditions.
Step 2: Go or no go?
The nature of the activities in this step is mixed. Developing ideas and criteria are
cyclical activities, that takes place at time-box intervals (NPD Sprint); drop criteria is a
unique activity, take 3 competitors – a unique activity, too. Like in previous step,
communication and current information about the status of activities are important.
Step 3: Testing the Concept
For all administrative activities and checking the marketing efficiency the suggested
type is a unique activities. Identification of consumer’s stage of knowledge of product
is recommended as a cyclical activity (NPD Sprint).
Step 4: Business Analysis
All activities contained in this step are cyclical (NPD Sprint). Step by step the NPD
Team goes toward the full knowledge concerning the new product from business point
of view.
Step 5: Marketing Tests
The activities in the frame of this step consist in elaboration different versions and
testing them. Recommended is their cyclical character (NPD Sprint).
Step 6: Technicalities + Product Development
There are three principal stakeholders in this step: production, marketing and finance
department. Two first activities are prepare respective plans (production and market-
ing); suggested classification of them - cyclical activities (NPD Sprint), because each
plan’s elaboration is cyclical. The third activity - provide the finance - is typically a
unique one.
Step 7: Commercialize
At this stage, there are two activities: keeping the distribution pipelines loaded with
products and maintain the product’s name in the mind of customers. Both are con-
tinuous, in which new facts appear, which is why it is recommended as a cyclical
activities, carried out in time-box iterations (NPD Sprint).
Step 8: Post Launch Review and Perfect Pricing
Reviewing process and perfecting pricing are both continuous activities. So, in this
proposal, they should be organised as cyclical (NPD Sprint).
The analysis of activities showed that in the project a significant part of the
activities are cyclical activities, repeated at certain intervals of time. This is due to NPD
steps character (innovating, planning, improving and so on). Unique activities appear
as a result of respective cyclical actions. Most of the actions described in this study
should be carried out continuously, at the agreed time intervals (NPD Sprints), until the
project’s goal is achieved. The result of the analysis is presented in the table below.
198 J. Betta et al.
Table 1. Activities in NPD process

Cyclic activities Unique activities
Step 1., Continuous improvement Any
Step2., Developing ideas and criteria Drop criteria
Take 3 competitors
Step 3., Identification of consumer’s knowledge Administrative activities
Checking the marketing efficiency
Step 4., Developing knowledge about new product Any
Step 5., Elaboration different, versions and testing them Any
Step 6., Prepare respective plans Provide the finance
Step 7., Keeping the distribution, Any
maintain the product’s name
Step 8., Reviewing process, perfecting pricing Any
Development and management of the NPD Product Backlog should remain within
the competence of the NPD Product Owner. In the case of the NPD Sprint Backlog, its
development and management is entrusted to the NPD Team.
During the project realization, it is proposed to adopt an iterative approach (NPD
Sprint). The duration of the iteration should be constant, adapted to the speed of
changing operating conditions. In the course of a single iteration, actions from the NPD
Sprint Backlog are carried out. The NPD Sprint Backlog includes cyclical and/or
unique activities selected by NPD Team. Due to the specificity of the situation, it is
recommended that the workload in the NPD Sprint Backlog is <60–80% of the planned
time. This is due to the fact that in emergency situations (e.g. a sudden new version
appears) it may be necessary to add new tests to be made. It is worth having a “time
buffer” that will be a response to sudden events. In the absence of unforeseen actions,
the NPD Sprint will be closed earlier.
It is proposed to use some meetings that will allow inspection and adaptation to
changing conditions. These are meetings defined on the basis of the Scrum method, [7,
8]:
– Planning NPD Sprint - meeting aimed at analysing the situation, determining the
purpose of iteration and creating the NPD Sprint Backlog. The meeting recom-
mends taking part in NPD Scrum Master, NPD Product Owner, NPD Team and key
project stakeholders (e.g. representatives of potential customers). The main product
of the meeting will be the NPD Sprint Backlog, which will define the work nec-
essary for the implementation of a NPD Sprint.
– Daily NPD Meeting – daily, brief meeting for the assessment of what has been done
during last 24 h, what will be done in the next 24 h and the current assessment of
the situation. Meeting is intended for members of the NPD Team to ensure
inspection of undertaken activities and adaptation to changing conditions.
– NPD Sprint Review - a meeting at the end of iteration for the assessment of the
implemented activities and the assessment of the degree of implementation of the
project’s objective. The meeting recommends taking part in NPD Scrum Master,
NPD Product Owner, NPD Team and key project stakeholders (e.g. representatives
of potential customers). During the meeting, it is recommended to:
• discuss the current situation and how it will change;
• present of what went well, what problems were encountered and how these
problems were solved;
• present of the work done;
• if the need arises, a probable deadline for completing activities (project com-
pletion) is expected;
• a joint discussion of the next steps;
• revising time, budget.
– NPB Sprint Retrospective - meeting for planning improvements for the next iter-
ation, intended for the NPB Scrum Master and the NPB Team.
The above transfer of Scrum elements to NPD have been proposed for ten cyclical
activities in the process (Table 1). For five other activities, which are unique, Scrum is
not applicable. In such case, these activities will be performed once in the project.
These activities will occur in the event of certain circumstances (detected during the
monitoring of the situation). The decisions will be often take ‘ad hoc’. Kanban might
be used here, but this would require further research.
6 Conclusions, Recommendations
In Sects. 3 and 4 the basic principles of agile approach to project management have
been presented, as well as, particularly, of the Scrum methodology. The most important
of these two section is the universality of Scrum, confirmed by increasing number or its
successful applications in management of different kinds of project, [10–12]. That’s
why the authors see a strongly justified possibility to apply Scrum in NPD Process
management, which should improve its management and its final result, too. Further
research as well as experimental implementations are necessary.
This matter – introducing the principal elements of Scrum into NPG – was analysed
in the Sect. 5. The double nature of the NPD process activities has been shown. Most
of them have a cyclical character, such as it practiced in Scrum. The Scrum elements in
the studied process: Roles, Events and Artefacts have been presented in details.
The main recommendation is directed toward the specialists of the NPD process.
From several years, Scrum is applied successfully in other domains, like Construction
Industry, [13]. Another possibility of using Scrum in NPD process is offered by hybrid
methodology, [14]. Hybrid is not a new methodology, but a fusion of two old
methodologies – traditional and agile. It has been practiced by experienced project
managers for many years under different names. Recently the name “Hybrid Project
Management” has gained acceptance.
The authors encourage them to implement Scrum in their work. The results will
come quickly. Note that it is possible to apply Scrum little by little.
200 J. Betta et al.
References
1. 8 step process perfects product development. http://www.innovationcoach.com/2013/05/8-
step-process-perfects-product-development/. Accessed 11 June 2018
2. Bessad, J., Francis, D.: Implementing the new product development process. Technovation
17(4), 189–222 (1997)
3. Schwaber, K., Sutherland, J.: Scrum guide. The definitive guide to scrum: the rules of the
game. http://www.scrumguides.org. Accessed 11 June 2018
4. Manifesto for agile software development. http://agilemanifesto.org. Accessed 10 Jan 2017
5. Matuszek, J., Seneta, T.: Algorithmization of new product process implementation in terms
of mass production, Mechanik Nr 7/2016
6. Betta, J., Skomra, A.: How to make a traditional project management more agile – an
educational need. In: Proceedings of INTED, Walence (2018)
7. Betta, J., Skomra, A.: Project risk management in the SAMI model. In: Proceedings of
Katowice University of Economics (2018, accepted to be published)
8. Betta, J., Skomra, A.: Agile crisis management. J. Sci. Mil. Acad. Land Forces (2018,
accepted to be published)
9. Betta, J., Drosio, S., Kuchta, D., Stanek, S., Skomra, A.: Agile approach in crisis
management - a case study of the anti-outbreak activities preventing an epidemic crisis.
In: 39th International Conference on Information Systems Architecture and Technology,
ISAT 2018, pp. 185–195. Springer (2019)
10. Leyrie, Ch., Boivin, S., Betta, J., Skomra, A.: Réflexions autour des contributions possibles
du croisement des approches agiles et de la co-construction, conférence au Colloque, 417 –
86e, congrès de l’ACFAS – Mardi 8 mai 2018, Quebec (Canada) (2018)
11. Sutherland, J, Hegarty, C.: Scrum in church. In: Agile Conference 2009 (2009)
12. Gandomani, T., et al.: Agile transformation: what is it about? In: 8th Malaysian Software
Engineering Conference (MySEC) (2014)
13. Streule, T., et al.: Implementation of scrum in the construction industry. Procedia Eng. 164,
269–276 (2016)
14. https://www.binfire.com/hybrid-project-management-manifesto/
Smart Industry Solutions
Drivers Impacting Cobots Adoption
in Manufacturing Context: A Qualitative Study
Ana C. Simões1(&), António Lucas Soares1,2, and Ana C. Barros1

1
INESC TEC, Rua Dr. Roberto Frias, 4200-465 Porto, Portugal
ana.c.simoes@inesctec.pt
2
Faculty of Engineering, University of Porto, Porto, Portugal
Abstract. Today’s manufacturing environment is increasingly pressured to

higher flexibility induced by uncertain production volumes as well as uncertain
product lifetime. A way to improve productivity in a flexible production system
is by using a safe and flexible cooperation between robot and operator.
Therefore, manufacturing companies are experiencing an increase in human-
robot interactions and in the use of collaborative robots (cobots). To make full
use of cobots, it is essential to understand the drivers for their adoption as well
as how these drivers are aligned with the companies’ strategic objectives. By
means of in-depth interviews in six companies in Portugal and France, this study
provides a comprehensive understanding of the drivers that influence the intent
to adopt, or the effective adoption, of cobots and the alignment of these drivers
with the strategic objectives of the company. Empirical results reveal “opera-
tional efficiency” and “ergonomics and human factors” concerns as important
drivers in the adoption intent. In terms of strategic objectives, it was found that
drivers are aligned with productivity and flexibility improvement as well as
quality improvement strategic objectives. Understanding these drivers can help
in motivating manufacturing companies to adopt cobots, in facilitating their
adoption, and in reaping the benefits of this technology.
Keywords: Cobots Adoption drivers Strategy Qualitative study
1 Introduction
In the last decades, the manufacturing ecosystem lived an unprecedented evolution of

disruptive technologies, which has helped companies dealing with competitive pres-
sures coming from product complexity, shorter product life cycles, change in demand
towards more customized product, rapid time-to-market, cost pressure, increased
international competition, etc. [1, 2]. To handle with these challenges, organizations
continually seek for flexibility and adaptability of manufacturing processes, which
require a close cooperation between the worker and the automated system [3].
Therefore, competitive levels in the manufacturing industry organizations are achieved
investing in new technical developments (e.g. advances in robotics, artificial intelli-
gence, and machine learning). The current flexible automation techniques, including
flexible manufacturing systems (FMS), and reconfigurable manufacturing systems
(RMS), lead to a recent trend in robotics that make use of the new generations of robots

https://doi.org/10.1007/978-3-030-18715-6_17
204 A. C. Simões et al.
with the capability to directly assist humans [4]. Flexibility and adaptability of man-
ufacturing processes require a close linkage between the worker (human) and the
automated system.
Cobots are an advanced manufacturing technology primarily characterized by the
fact that the robots’ and workers’ work zones overlap, creating a common workspace,
without safety fencing or barriers in the manufacturing floor [5, 6]. According to a
recent study carried out in German companies [6], humans and robots primarily work
alongside each other in a form of coexistence (both interaction partners may have tasks
to perform at the same time in the shared workspace, but they do not work simulta-
neously on the same product or component), an arrangement in which the new tech-
nology is very reliable. However, collaborative applications (human and robot working
simultaneously on the same product or component) are virtually non-existent in pro-
duction facilities at that time [6]. Cobots are designed for the assembly line worker to
reduce ergonomics concerns that arise due to on-the-job physical and cognitive load-
ing, while improving safety, quality, productivity and flexibility in plant operations [7,
8]. Then, a safe and flexible cooperation between machines and humans can be
achieved, because it matches the strength and the efficiency of robots with the high
degree of dexterity and the cognitive capabilities of humans achieving better produc-
tivity at the most flexible overall system [4, 9]. Furthermore, reducing the limits of
robotic automation to allow a distribution of tasks between humans and robots enables
the applicability of robots for industrial production. Humans and robots each take on
the tasks for which they are best-suited, with frequent interaction and shared procedures
[10]. However, the human must be kept in the centre of decision process in the
production process for highly flexible assembly due to human’s cognitive and senso-
motoric advantage [4].
The interaction between human and robot improves complex assembly processes,
particularly when a worker can guide a robot and the robot provides power support to
the worker. The support provided by the robot to the worker in manufacturing pro-
cesses are related to difficult, monotonous or physically demanding tasks [3]. Advances
in sensors, actuators and data processing enable to improve the degree of assistance [3,
10]. The human should be part of production processes when he is needed, otherwise
he can concentrate on other tasks to improve the overall system performance [4].
Decisions regarding automation in organizations are classified as process (tech-
nological) decisions. Nevertheless, this is a technical perspective since it does not take
into consideration the integration of humans and technology and fails to explain the
decisions for the selection of technological investments that support the company
strategy [11]. With the growing implementation of cobots in the manufacturing pro-
cesses, and thus the integration of humans and technology in order to achieve more
efficient and robust production systems, the alignment between the automation deci-
sions and the company’ strategic objectives are critical for the company success [12].
The objective of this paper is to understand the drivers for the cobots adoption as well
as how these drivers are aligned with the companies’ strategic objectives.
Drivers Impacting Cobots Adoption in Manufacturing Context 205
2 Problem
In light of the benefits and challenges of cobots, it is important to understand the drivers
that influence the adoption of cobots in manufacturing companies, in order to facilitate
a successful implementation of this technology. Current research within the field of
human-robot interactions is suggesting that there are several challenges in the imple-
mentation of cobots (acceptance, security, etc.) that require further research [13]. This
paper is contributing with empirical evidence to reduce this literature gap by
researching important drivers that influence the adoption of cobots and the alignment of
these drivers with the strategic objectives of the company. Hence, the research question
for this paper is: What are the drivers that influence the intent to adopt, or the adoption,
of cobots by manufacturing organizations, and how are they aligned with the com-
panies’ strategic objectives?
3 Method
The research reported in this paper is exploratory in nature and utilizes qualitative
evidence. The purpose of this study is to identify and understand the drivers to adopt
cobots in factories of multinational companies installed in Portugal and France. Given
the novelty of this technology to manufacturing companies, this study considers that
drivers conducting to cobots adoption can be better understood by examining the
interpretations of managers [14, 15] in different function and industrial sectors, justi-
fying a qualitative research approach.
Qualitative data were collected, providing rich insights for exploring, identifying,
and understanding viewpoints regarding the adoption drivers, as well as offering the
opportunity to understand deeply the topic and clarify misunderstanding aspects [16,
17]. Six companies were approached and agreed to participate to participate in the
study. A total of thirteen interviews were conducted between February and May 2018,
and had duration between 30 and 70 min. Each interview was conducted in the Por-
tuguese language, with the exception of one (C3I1) that was conducted in English
language. The open-ended style of the interviews allowed the respondents to describe
their opinions, experiences and ideas freely, without being limited to predefined con-
cepts or models. Interviewees’ characterization is presented in Table 1.
The audio files from the interviews were transcribed and MAXQDA (version 11.0)
qualitative analysis software was used to analyze the interviews. A thematic and
theory-driven analysis was adopted [18]. First, a subset of interview transcripts were
analyzed (a) to develop an initial coding scheme for the pre-defined interview topics
and (b) to identify topics in the transcripts that were not specifically queried by the
interview guide [19]. After that, a structural coding was used to identify unanticipated
topics that occur in the interview [18–20].
Construct validity has been addressed, first by using multiple sources of informa-
tion [21]. While interviews constitute the primary source of information, some support
information was collected in field notes (collected during the field visits to the facili-
ties). In addition, the researchers themselves identified additional supporting docu-
mentation including materials located at the websites of the informants’ organizations
or in publications associated with the industry. Second, considering different per-

spectives constitutes an important type of triangulation of qualitative information
sources [16]. Different perspectives were provided by the different key informants (see
third column in Table 1). Finally, the chain of evidence, tracing the conclusions to the
interview summary and to the interview transcripts was also maintained. According to
Yin [21] these enhance the construct validity as well as the reliability of the research,
thereby boosting its overall quality.
Table 1. Interviewees’ characterization

ID Industrial Function Duration
sector (min)
C1I1 Security Industrial eng., process engineer and internal logistics 38
system group leader; I4.0 project manager
C2I1 Package Divisional industrial director – packaging division 60
C3I1 Automotive Advanced manufacturing department in charge of 56
powertrain factory of future
C4I1 Automotive West operations manager 31
C4I2 components Process engineering manager 36
C4I3 Engineering and innovation director 41
C4I4 Maintenance supervisor 30
C4I5 Operations director 43
C5I1 Automotive Lean management - continuous improvement 52
C5I2 Assistant manager - maintenance & facility Written
management form
C6I1 Automotive Managing director 65
C6I2 components Maintenance supervisor 32
C6I3 Process engineer 70
4 Analysis and Results
In the following sections the results of the interviews are presented. First, describing
the companies in relation to the main aspects of this study, and then the drivers and
their relationships with companies’ strategic objectives. The data gathered from the
interviews and by other sources of evidence provided the necessary information to
understand the drivers that influence the intent to adopt, or the effective adoption, of
cobots as well as to understand the alignment of these drivers with the strategic
objectives of the company.
4.1 Companies Description

Table 2 characterizes the companies for the purpose of this study.
Table 2. Companies’ characterization

Company Plant Products Headquarter Employees Turnover Classification Cobots
location (FTE) 2017 (M€) (FTE > 249) adoption
C1 Portugal Products related Germany 700 102 Large No
to security
systems
C2 Portugal Plastic packages Portugal 950 142 Large No
C3 France Cars France 170 000 660 000 Large No
C4 Portugal Automotive Portugal 3 206 396 Large No
plastic
components
C5 Portugal Trucks Japan 426 214 Large Yes
C6 Portugal Automotive Germany 422 110 Large No
components
In the last column, information related to the current situation related to cobots
adoption: yes, if the company have cobots in its production process; no, if the company
don’t have yet cobots in its production process but intend to have them in the near
future.
4.2 Drivers
During data analysis several drivers that influence the intent to adopt, or the effective
adoption, of cobots were found in the empirical analysis. In the next sections these
drivers are presented. The drivers will be presented and described considering, among
other aspects, the frequency that participants referred this driver during the interviews.
For the purpose of this study, a driver that was mentioned by all the participants is
considered more important than one that was only mentioned, for example, by two
participants.
4.2.1 Operational Efficiency

Operational efficiency was the driver more frequently mentioned by the interviewees
and has been mentioned by all the 13 participants, which gives to this driver a great
importance. The analysis of the empirical data allows for a detailed description of this
important driver. Participants referred productivity improvement (efficiency improve-
ment), quality improvement, flexibility improvement (efficient customization), work-
space gains and complexity reduction as aspects (or “sub-drivers”) to be considered in
operational efficiency. Although, all these “sub-drivers” were mentioned in this study,
the importance given by the participants was not the same for all. Thus, the sub-drivers
mentioned with more frequency were, in this order, the productivity improvement
(efficiency improvement), the quality improvement and the flexibility improvement
(efficient customization).
According to participants, the adoption of cobots leads to cost reduction, mainly

due to wage cost reduction, which is viewed as an important contribute to improve
efficiency. Additionally, the consistency improvement (through task repetition) is also
considered as having an important contribution to improve quality (less errors) and
consequently the productivity.
“We can say that our customer, in this case the holding company, has the driver Productivity.
So, we [the subsidiary company] have to consider certain KPIs, or certain objectives, to
accomplish. Therefore, this [productivity] is our driver.” (C5I1)
The size of the cobots as well as the possibility that cobots have to execute tasks
without fences, improving the available workspace, was also referred as a sub-driver
that influences positively the operational efficiency. However, safety aspects related to
the interaction between human workers and robots, distinct of this technology, were
reflected as concerns of the participants of this study (C1I1, C3I1, C4I2, C5I1, C6I2
and C6I3). Although the numerous advantages of the collaborative robots, the partic-
ipants C2I1, C3I1 C4I1, C4I2, C5I1 and C6I2 were aware of the limitations of this
technology in terms of velocity (to execute the tasks) and impact. In order to achieve
the safety requirements, the cobots execute the tasks with a low velocity (when
compared to fenced traditional robots), which has a negative implication on produc-
tivity. According to the participants, risk assessment procedures and the application of
safety and medical/biomechanical requirements must be considered in order to avoid
injuries to the human workers (to keep any risk of injury to a low and acceptable level).
4.2.2 Ergonomics and Human Factors

Ergonomic and human factors concerns were the second most mentioned driver in this
study. Almost all participants were unanimous referring that the adoption (or the intent)
to adopt is firstly driven by ergonomic and human factors concerns. Regarding this
driver, participants mentioned that the adoption of this technology would avoid
assigning to human worker dangerous, repetitive and boring tasks. Additionally, cobots
will support human worker giving power assistance and avoiding postural problems.
“I take it for granted that them [cobots] are coming to help, in repetitive operations, that in the
ergonomic point of view are more difficult, or have more propensities to cause muscu-
lar and skeletal disorders etc., etc., I see them [cobots], in fact, as an asset.” (C6I1).
Ergonomic improvements have strong links with human resources sustainability,

according to the results of our study. The interviewee C5I1 mentioned that ergonomic
optimisation is related to the advanced age of some the human workers in production
lines:
“visual difficulties increase with age, they will also have locomotion difficulties, which means
that these cobots, can also be a positive asset.” (C6I3)
Other participants (C5I1, C4I2, C2I1) share the same concerns as C6I3. These
participants look at these senior human workers as a valuable asset for the company,
since they have the knowledge about the production process that young people don’t
have. The adoption of cobots can be driven by the desire to maintain that senior
workers as much time as possible giving them the best work conditions. These
participants see in the cobots the support that these senior workers need to execute their
tasks better and with quality.
For participant C2I1, the cobot adoption is driven by the need to balance the
demand and supply labour needs in certain geographic regions. Additionally, the
expectations and experience of younger and older human workers also need to be
considered.
“On the one hand, people who come to the plant [for the first time] don’t adapt themselves to
the conditions that plant offer, [the plant] is unattractive for them, on the other hand the people
in the plant have difficulties in the adoption of new technologies and to the interfaces.” (C2I1)
4.2.3 Industrial Innovation Context

The participants refer innovation as an important driver to the adoption of cobots. Eight
participants mentioned this driver. For C5I1 the possibility to be the first company in
the group to implement in the production process a cobot is considered an important
driver, since they want to be considered an innovative company in their group. C6I3
also referred the importance of implement new technologies in the production process,
since according to him can be a competitive advantage. C1I1, C6I1 and C3I1 also
shared the same opinion.
“Another external factor is the need to show [for those outside the company] that we didn’t
freeze in time regarding the technological development, that we search for the best for what we
want to produce, and at the same time is the best to our customers.” (C6I1)
Other participants see the industrial context, in this case the automotive industry, as
a driver to adopt new technologies. The technology progress as well as the investments
on the development of new technologies made by companies in this sector is con-
sidered by C4I4, C3I1 and C2I1 as one of the most relevant drivers.
4.3 Alignment with Company Strategy

The empirical results show that the companies follow three main strategic objectives
for the participants companies: improve productivity, improve quality and promote
innovation.
Almost all companies’ participants in this study referred productivity and efficiency
as the most important strategic objective. It is notorious the alignment of this objective
with the driver “Operational efficiency”. Several aspects mentioned related to this
driver contribute to improve the efficiency and productivity, such as improving flexi-
bility (batch sizes and cobot functions), reducing labour costs and reducing non-quality
costs.
As a strategic objective, the participants view quality improvement in two per-
spectives: quality of working life (satisfaction) of the workers and the final quality of
the product. Therefore, the driver “ergonomics and human factors” is aligned with both
objectives, since it improves quality in the perspective of the satisfaction of the human
workers as well as improves the quality of products themselves.
“The other strategy is to guarantee that the operator has other [better] quality is the work
place, which he doesn’t have today.” (C6I3)
Participants from companies C2, C4, C5 and C6 put great emphasis on strategic
objectives related to quality of working life and these objectives are very aligned with
the driver “ergonomics and human factors”.
Additionally, the quality improvement by reducing non-quality products (with
direct implications on cost) as well as a better satisfaction of the customer is mentioned
in this study by ten participants, as an important strategic objective and is aligned with
operational efficiency driver.
“(…) extensive assemblies, where the probability to fail is considerable for the operator, the
probability to fail in assembling a small part is high, the existence of a cobot can eliminate this
risk of fail, it is a win-win situation.” (C1I1)
Finally, promoting innovation is the third strategic objective mentioned in this

study. According to the results, this objective is aligned with the previous mentioned
driver with the same name, “Industrial innovation context”. For all companies partic-
ipating in this study the implementation of new technologies (industry 4.0 technolo-
gies) in their production processes is part of the strategy.
The adoption of this kind of technology meets the group strategy, there is a group strategy
regarding the adoption of Industry 4.0 technologies. So, this means that there is clearly a
vision.” (C6I3)
C4I2 mentioned that the innovation is a strategic goal always present, and the
adoption of this new technology, namely automation technology, is the natural path of
the company. For C4I3 the innovation, is indirectly linked to the results (in terms of
processes efficiency), since new technologies when adopted improved these results.
5 Conclusion
Cobots are considered new technologies with important benefits for companies. The
results of this study reveal that operational efficiency, ergonomics and human factors,
and industrial innovation context are relevant drivers with influence on the intent to
adopt, or the adoption, of cobots. It was possible to conclude that there is an alignment
of these drivers with the strategic objectives of the company, since these drivers are
reflected in the companies’ strategic objectives. However, challenges related to their
implementation, are reflected in some concerns expressed by company managers,
mainly the safety aspect. The priority is to guarantee worker safety while simultane-
ously avoiding intermittent disruption of the robot’s work. Additionally, the acceptance
of this technology by human workers, that have to share the same workspace with
cobots, was also pointed as a challenge by the participants. Clearly, the study reported
in this paper is based on the Portuguese and French contexts, and thus, it may not be
readily generalizable beyond this study. Further research is required, both in Portugal
and in other countries and industries to ensure generalizability. Future research should
consolidate these results with other companies in different industries and in different
countries.
Acknowledgments. This work was supported by Portugal 2020 project “DM4Manufacturing -

Aligning manufacturing decision making with advanced manufacturing technologies”, POCI-01-
0145-FEDER-016418, financed by UE/FEDER through the program COMPETE2020 and by the
Project “TEC4Growth - Pervasive Intelligence, Enhancers and Proofs of Concept with Industrial
Impact”, NORTE-01-0145-FEDER-000020, financed by the North Portugal Regional Opera-
tional Programme (NORTE 2020), under the PORTUGAL 2020 Partnership Agreement, and
through the European Regional Development Fund (ERDF).
References
1. Brettel, M., et al.: How virtualization, decentralization and network building change the
manufacturing landscape: an industry 4.0 perspective. Int. J. Mech. Ind. Sci. Eng. 8(1), 37–
44 (2014)
2. De Carolis, A., Tavola, G., Taisch, M.: Cyber-physical systems in manufacturing: future
trends and research priorities. In: 21st Summer School Francesco Turco 2016, Italy (2016)
3. Krüger, J., Lien, T.K., Verl, A.: Cooperation of human and machines in assembly lines.
CIRP Ann. - Manuf. Technol. 58(2), 628–6469 (2009)
4. Lenz, C., Knoll. A.: Mechanisms and capabilities for human robot collaboration. In: IEEE
International Symposium on Robot and Human Interactive Communication, Edinburgh, UK.
IEEE (2014)
5. Colgate, J.E., et al.: Cobots: robots for collaboration with human operators. In: Proceedings
of the International Mechanical Engineering Congress and Exhibition, Atlanta, USA (1996)
6. Bauer, W., et al.: Lightweight robots in manual assembly best to start simply. Frauenhofer-
Institut für Arbeitswirtschaft und Organisation IAO, Stuttgart. (2016)
7. Akella, P., et al.: Cobots for the automobile assembly line. In: International Conference on
Robotics and Automation, Detroit, MI, USA. IEEE (1999)
8. Calitz, A.P., Poisat, P., Cullen, M.: The future African workplace: the use of collaborative
robots in manufacturing. SA J. Hum. Resour. Manag. 15(1), 1–11 (2017)
9. Hägele, M., Schaaf, W., Helms, E.: Robot assistants at manual workplaces: effective co-
operation and safety aspects. In: 33rd ISR (International Symposium on Robotics) (2002)
10. Fryman, J., Matthias, B.: Safety of industrial robots: from conventional to collaborative
applications. In: Robotics, Munich, German. VDE (2012)
11. Hill, T.: Manufacturing Strategy: Text and Cases, 2nd edn. Palgrave Macmillan, Basingstoke
(2000)
12. Lindström, V., Winroth, M.: Aligning manufacturing strategy and levels of automation: a
case study. J. Eng. Tech. Manag. 27(3–4), 148–159 (2010)
13. Djuric, A.M., Urbanic, R., Rickli, J.: A framework for collaborative robot (CoBot)
integration in advanced manufacturing systems. SAE Int. J. Mater. Manuf. 9(2), 457–464
(2016)
14. Gray, D.E.: Doing Research in the Real World, 2nd edn. Sage, Thousand Oaks (2009)
15. Flick, U., von Kardoff, E., Steinke, I.: What is qualitative research? An introduction to the
field. In: Flick, U., von Kardoff, E., Steinke, I. (eds.) A Companion to Qualitative Research.
Sage, Thousand Oaks (2004)
16. Patton, M.Q.: Qualitative Research & Evaluation Methods, 3rd edn. Sage Publications,
Incorporated, Thousand Oaks (2002)
17. Mason, J.: Qualitative Researching, 2nd edn. Sage, Thousand Oaks (2002)
18. Namey, E., et al.: Data reduction techniques for large qualitative data sets. In: Guest, G.,
MacQueen, K. (eds.) Handbook for Team-Based Qualitative Research, pp. 137–162.
AltaMira Press, Lanham (2008)
19. MacQueen, K.M., et al.: Codebook development for team-based qualitative analysis. Cult.
Anthropol. Methods 10(2), 31–36 (1998)
20. Saldaña, J., The Coding Manual for Qualitative Researchers (2009)
21. Yin, R.K.: Case Study Research: Design and Methods, 4th edn. SAGE Publications, London
(2009)
Control of the 6-Axis Robot Using
a Brain-Computer Interface Based on Steady
State Visually Evoked Potential (SSVEP)
Arkadiusz Kubacki(&) and Andrzej Milecki(&)
Faculty of Mechanical Engineering and Management, Division of Mechatronics

Devices, Poznan University of Technology, Piotrowo 3, 61-138 Poznań, Poland
{arkadiusz.kubacki,andrzej.milecki}@put.poznan.pl
Abstract. This paper presents the research on the possibility of control of the 6-
axis robot using a brain-computer interface (BCI) based on Steady State Evoked
Potential (SSVEP) signals. In the first paragraph, general information about
brain-computer communication are presented. In the next paragraph the used in
investigations equipment is described. The recorded SSVEP signals and graphs
obtained in initial tests are shown. In the last part, the experimental procedures
and results of research are described.
Keywords: Steady State Visually Evoked Potential SSVEP

Brain-computer interface Electroencephalography EEG Robot control
1 Introduction
To operate of production devices, different Human Machine Interfaces are used, like
LED indicators and mechanical Switches (a leading HMI for industrial applications),
capacitive touch elements, industrial control panels with displays and keyboards, touch
screens, joysticks, haptic devices etc. The innovative systems use bio signals coming
straight from the human body or brain. These signals change their shape as a result of
change of the physiological state of a human. Bio signals can be used to provide input
signal to different mechatronic devices, thus to control them [1]. Currently, brain-
computer interfaces are the most promising systems capable of communication with a
computer for bedridden patients. To monitor brain activity following solutions can be
used [2]:
• Electroencefalography (EEG),
• Positron emission tomography (PET),
• Functional magnetic resonance imaging (fMRI),
• Magnetoencephalography (MEG).
At the moment, EEG is developing rapidly thanks to decreasing price of the headsets
[3]. For this reason, most of the brain-computer interface solutions are based on these
devices and appropriate to them methods are used. Electroencephalography is a non-
invasive way of obtaining information about the activity of the brain through electrodes
placed on the scalp of the human. The most common system of arrangement of the

https://doi.org/10.1007/978-3-030-18715-6_18
214 A. Kubacki and A. Milecki
electrodes location on the human head, is a so called “10–20” system [4]. This system
is an internationally recognized method to describe and apply the location of scalp
electrodes in the context of an EEG test or experiment. This method was developed to
ensure standardized reproducibility so that a subject’s studies could be compared over
time and subjects could be compared to each other. This system is based on the
relationship between the location of an electrode and the underlying area of cerebral
cortex. The “10” and “20” refer to the fact that the actual distances between adjacent
electrodes are either 10% or 20% of the total front–back or right–left distance of the
skull. The electrodes placement in this system is shown in Fig. 1. Each electrode site is
labeled with a letter and a number. The letter refers to the area of brain underlying the
electrode e.g. F - Frontal lobe and T - Temporal lobe. Even numbers denote the right
side of the head and odd numbers the left side of the head.
Currently, in the brain-machine interfaces two methods to measure brain activity
are used: passive and active. Passive methods need external stimuli to induce brain
activity. In this group are two method – SSVEP and P300. Potentials in Steady State
Visually Evoked Potential (SSVEP) method are most visible in the area of the electrode
Oz [3, 5–7]. Potentials in P300 method are most visible in the area of the electrode Pz
[8–10]. In active methods, human brain activity is caused through adequate imaginary.
In this group is only one method – ERD/ERS. Signals in this method are most visible in
the area of the electrodes Cz, C3 and C4 [11–14].
Fig. 1. Electrode placement in accordance with the “10–20” system [4]
2 State of Art
Currently, scientists in the world are designing brain-computer interfaces using both
active and passive methods.
Control of the 6-Axis Robot Using a Brain-Computer Interface 215
In the article [11], the authors describe their brain-computer interface for control-
ling a wheelchair. The control was carried out by means of issued commands. To drive
straight, the driver had to imagine raising both feet to the top. The change of direction
took place through the image of raising the left or right hand. The authors used a
support vector machine (SVM) to classify signals originating from electroen-
cephalography. The driver’s task was to reach the destination by driving on an figure-8
course. The success rate for the last attempts was over 90% of trips. Three men aged 27
to 33 participated in the study.
The authors of the article [12] used ERD/ERS for control. In this article, the
subjects were controlled by a virtual helicopter. The subjects were tasked with flying
through the rings placed in a virtual room. The authors have used for this purpose
commands based on the imagery of turning left and right and lift and descent. Rotation
of the helicopter was carried out via the left/right command. Imagining the left
movement, the helicopter turned left. Similarly, when moving to the right. The vertical
position is determined by imagining the up/down command. The authors created three
scenarios. In the first scenario, the circles were set in a fixed position. After every point
the helicopter position was reset. In the second scenario, the position of the rings was
still constant. This time the helicopter’s position was not reset. In the last scenario, the
position and orientation of the rings was chosen randomly. Four people participated in
the study.
Another method is described in the article [8]. The authors utilizing the evoked
potential P300 created their own speller. With the proper sequence of flashing letters,
they were focused on. The authors used Linear Discriminant Analysis (LDA) and Fisher
Linear Discriminant (FLD) to classify systems derived from electroencephalography.
Using evoked potentials and stimulating subjects by stimuli of frequencies from
6.66 to 8 Hz authors in the article [6] created a speller based on Steady-State Visually
Evoked Potential. Comparing this speller with the article [8], the authors divided the
alphabet into three subgroups, thanks to which they significantly reduced the time
needed to enter the character. After careful observations a particular sub-group have
been selected and the letters in it were divided into further sub-groups, and so on until
the tested person would not select a single letter. In this article, authors used a gen-
erative approach based on the principal component analysis (PCA). Eight healthy
subjects participated in this study. The average age is 30 years, with a standard
deviation of 3.66. Only one female subject was present. Three people need vision
correction and wore their glasses. The subjects were asked to write the words BCI,
BRAIN, CERVEAU, SSVEP, BRAIN_COMPUTER_INTERFACE, the first name of
the subject and one word chosen by the subject. The average word entry time was
10.65 min.
In the article [7] they presented the electrical prosthesis controlled by the brain-
computer interface based on Steady-State Visually Evoked Potential (SSVEP).
Blinking fields had a rate of 6 Hz for the command turn hand left, 7 Hz for the
command turn hand to the right, 8 Hz forth command open hand, 13 Hz for the
command close hand. Four healthy volunteers aged from 18 to 29 years old (two males,
and two females) participated in this study. All had normal or corrected to normal
vision. Stimulation unit was situated about 0.8 m in front of tested person. The authors
obtained classification rates from 74 to 88%.
Article [3] shows the possibility of SSVEP to control a virtual car. Authors used
two HUD screens. The first HUD displayed images of frequencies 12 Hz. The second
displayed images of frequencies 13 Hz. Respondents get around previously designed
route. The authors used Linear Discriminant Analysis (LDA) to classify systems
derived from electroencephalography. Four people participated in the study. Three men
and one woman. A mean age of 25. They obtained satisfactory results for vehicle
speeds from 7.2 to 10.8 km/h.
3 Research Equipment
The main assumption was to build a brain-computer interface based on SSVEP to control
an industrial robot. Thanks to such a system, simple handling operations can be per-
formed. The presented in this paper study is conducted with the use of a neuroheadset
called Emotiv EPOC+ which is connected with a computer via Bluetooth (Fig. 2b). This
device consists of 14 EEG channels + 2 reference electrodes, which are placed in
accordance with the “10–20” system. Emotiv EPOC+ device uses one 16-bit ADC with a
sampling rate of 256 samples per second (2048 Hz internal). This device receives brain
waves in a bandwidth of 0.16 to 43 Hz which means that all five main brain rhythms
could be recognized [17]. These frequency are associated with several states of mind:
• Delta waves (1–3 Hz): Deepest meditation and deep sleep
• Theta waves (4–7 Hz): Normal sleeping and normal meditation
• Alpha waves (8–12 Hz): Relaxation/reflection
• Beta waves (13–30 Hz): Active thinking, focus, hi alert, anxiety
• Gamma waves (31–50 Hz): Conscious perception
In order to generate stimuli signals to the user authors have built a special blinking
lights with dimensions of 45 45 45 mm. Inside it, there is a 3 W Power RGB
LED (Fig. 2a). Thanks to this there was possible to flash a light in different colors and
with different frequencies. To control of this lamp a module based on Atmega 328P
microcontroller is developed, which allowed to set the diode frequency between 5 and
30 Hz and the brightness of the diode in the range of 0 to 100%. The intervals between
glare phases are random and range from 5 s to 10 s.
Fig. 2. Equipment used in tests: blinking module (a), Emotiv EPOC+ [7] (b)
4 Research and Results
Steady-State Visually Evoked Potential (SSVEP) is a method used in brain-computer

interface consists of a periodic measurement of evoked potentials generated by human
as a response of repeated visual stimulation like blinking light. Frequency of blinking is
usually more than 6 Hz [15]. Flashing light such as LED on which markers are flashing
or monitors with blinking chessboard patterns are commonly used as stimulating ele-
ments [16]. During the stimulation in EEG signal additional signal with frequency
corresponding to the stimulation and their harmonics [6] appear. This method can be
used on the test person without prior training [17]. During initial testing the correctness
of the SSVEP method is verified. Figure 3 shows the Fast Fourier Transformation of
signal taken from electrode called Oz. In Fig. 3a, the FFT is shown while the LED is
not blinking. On the right the FFT of the same electrode output signal is shown when
white LED flashed with frequency 15 Hz.
a) b)
300 300
Amplitude [µV]
Amplitude [µV]
200 200
100 100
0 0
10 12 14 16 18 20 10 12 14 16 18 20
Frequency [Hz] Frequency [Hz]
Fig. 3. Electrode Oz output signal FFT characteristics for: no lamp (a), lamp flashing with
15 Hz (b)
In the first experiment, the response time on the external stimulus is checked for
different frequencies. For this purpose, a test algorithm is created in OpenVibe system
[18]. The scheme of the algorithm (program) is shown in Fig. 4.
4th order
Time based
Input signal Butterworth Channel Selector
epoching
Bandpass filter
DSP DSP
Signal average Output command
(x*x) log(1+x)
Fig. 4. Block diagram of signal processing during the experiments, made in OpenVibe system
In the first step, the input signal from electrode is filtered using 4th order band pass
filter. In the performed test, following frequencies are set: 15 Hz, 17 Hz and 19 Hz. In
the next block, the appropriate input channel for the test is selected. The responsible for
the vision part of the brain, is located on the back of the head. So, the Oz channel i.e.
electrode is selected. Various Time Based Epoching (TBE) blocks are tested in the
investigations using times of 1 s, 2 s, 3 s and 4 s. In the next step, the signal amplified,
averaged, its logarithm is calculated and finally outputted.
TBE 1s TBE 3s LED state

TBE 2s TBE 4s
3
EEG Power [log µV ]
2.5
2
1.5
0.5
0
0 5 10 15 20
t [s]
Fig. 5. SSVEP response graph on stimulus for different times of TBE block
The investigation has shown that the shorter the TBE is, the faster the response is.
Assuming a detection threshold of 1.5 lV2, the external stimulus response time is
approximately equal to the TBE time. However, the shape of the output signal should
be taken into account in detection process. In Fig. 5 the recorded SSVEP signals for
different TBE are shown. There is visible that the shorter the time is, in the output
signal more disturbing oscillations are. For further study TBE time was assumed to be
equal to 3 s. It is also worth to notice that the same time is necessary to detect the light
blinking and to detect that the lamp is off (not blinking).
In the second experiment, an approach to independently recognize three different
flashing frequencies of the white LED is made. The LED light is turned on three times
with 10 s intervals. The flash time is lasted also as 10 s. The flashing frequency was set
sequentially 15 Hz, 17 Hz and 19 Hz. The signal processing is performed according to
block diagram shown in Fig. 5. TBE time was set to 3 s.
15Hz 17Hz 19Hz

2.5
EEG Power [log µV2]

2
1.5
0.5
0
0 10 20 30 40 50
t [s]
Fig. 6. SSVEP response chart for flashing light at different frequencies
In all attempts, the system was able to correctly recognize the flashing frequency. In
Fig. 6 a examples of the EEG output signals recorded in this experiment is shown.
There is visible, that every LED blinking frequency can be clearly recognized, because
in every case the adequate output signal is higher than the threshold 1.5 lV2. In the
third experiment, the difference in the EEG power signal value is checked for various
flashing colors of the LED. The flashing frequency was 15 Hz. Filtering of the signal is
done using algorithm shown in Fig. 5. TBE time is set to 3 s. The LED was turned on
for 10 s. Intervals between turns are also equal to 10 s. In the investigations following
LED colors are used: red, yellow, blue, green, purple and orange. Also In every turn on
of the LED, colors sequence is changed for every attempt. The investigations have
shown that the strongest response is observed for red and yellow colors. The weakest
answer was noted for blue and orange colors. The blue color response is so small, that it
was impossible to determine if the lamp is blinking (Fig. 7).
The last experiment was concentrated on building a brain-computer interface to
control the 6-axis robot produced by Mitsubishi. Four flashing red lights blinking with
different frequencies are used to stimulation of the human operator, which EEG signals
are measured, recognized and used as robot’s input commands. For movement com-
mand in X+ the frequency 15 Hz is used. For other commands following frequencies
are use: 17 Hz for Z+ , 19 Hz for X− and 21 Hz for Z−.
The operator is supposed to run the robot effector from the starting point to the
target point and precipitate the ball located at the end point. In each test the user had
intended to move the robot arm about 450 mm in X axis and about 500 mm in Z-axis.
red blue yellow

green purple orange
3
EEG Power [log µV2]
2.5
1.5
0.5
0
0 20 40 60 80 100 120
t [s]
Fig. 7. SSVEP response chart for flashing light in various colors
500
a)
400
Z [mm]
300
200
100
0
0 50 100 150 200 250 300 350 400 450 500
X [mm]
X position [mm]
b) 400 X position BCI signal
200
0
0 10 20 30 40 50 60 70 80 90 100
t [s]
c)
Z position [mm]
400 Z position BCI signal
200
0
0 10 20 30 40 50 60 70 80 90 100
t [s]
Fig. 8. Control of the 6-axis robot: the robot tip displacement graph (a), the position of the robot
in X axis (b), the position of the robot in Z axis (c)
The test subject had 5 min to do this exercise. An exemplary motion of the robot tip
is shown in Fig. 8. Between 30 completed trials, 24 were successfully finished in the
appointed time. This gives a score of 80%. The best time was 93 s. A position of the
robot in the start and end positions is shown in Fig. 9.
Fig. 9. A position of the robot in the start (a) and end (b) positions
5 Conclusions
The investigations have shown that, it is possible to move the 6-axis robot using the
brain-computer interface based on SSVEP. 80% of tests were successful. The best time
was 93 s. With this method, a big hurdle is the fatigue of the eyes. Continuous looking
at the flashing point is not convenient. The authors plan to expand the system. They
want to add additional interfaces so that they create hybrid brain-computer interface
system.
Authors also confirmed that the color of the stimulus is important in the devel-
opment and effectiveness of this type of brain-computer interface. However, there is
visible that further tests should examine the use of other classifiers. Classifying the
signal as correct or not correct, with a threshold has some limitations. This important
disadvantage is the reaction time. The investigations have shown the inability to set a
shorter TBE block time. In subsequent studies, it was decided to use other classifiers
such as Linear Discriminant Analysis (LDA), Support Vector Machine (SVM) or
artificial neural networks.
Acknowledgments. The work described in this paper was funded from 02/22/DSPB/1434.
References
1. Kubacki, A., Milecki, A., Application of bio signals in the brain-device interfaces. In:
Automation 2017, pp. 108–119 (2017)
2. Wolpaw, J.R., Birbaumer, N., McFarland, D.J., Pfurtscheller, G., Vaughan, T.M.: Brain–
computer interfaces for communication and control. Clin. Neurophysiol. 113(6), 767–791
(2002)
3. Bi, L., Fan, X., Jie, K., Teng, T., Ding, H., Liu, Y.: Using a head-up display-based steady-
state visually evoked potential brain-computer interface to control a simulated vehicle. IEEE
Trans. Intell. Transp. Syst. 15(3), 959–966 (2014)
4. Varadan, V.K., Oh, S., Kwon, H., Hankins, P.: Wireless point-of-care diagnosis for sleep
disorder with dry nanowire electrodes. J. Nanotechnol. Eng. Med. 1(3), 031012 (2010)
5. Allison, B.Z., Brunner, C., Altstatter, C., Wagner, I.C., Grissmann, S., Neuper, C.: A hybrid
ERD/SSVEP BCI for continuous simultaneous two dimensional cursor control. J. Neurosci.
Methods 209(2), 299–307 (2012)
6. Cecotti, H.: A self-paced and calibration-less SSVEP-based brain- computer interface
speller. IEEE Trans. Neural Syst. Rehabil. Eng. 18(2), 127–133 (2010)
7. Muller-Putz, G.R., Pfurtscheller, G.: Control of an electrical prosthesis with an SSVEP-
based BCI. IEEE Trans. Biomed. Eng. 55(1), 361–364 (2008)
8. Elsawy, A.S., Eldawlatly, S., Taher, M., Aly, G.M.: Performance analysis of a principal
component analysis ensemble classifier for emotiv headset P300 spellers. In: 2014 36th
Annual International Conference of the IEEE Engineering (2014)
9. Elsawy, A.S., Eldawlatly, S., Taher, M., Aly, G.M.: A principal component analysis
ensemble classifier for P300 speller applications. In: 2013 8th International Symposium on
Image and Signal Processing and Analysis (ISPA), pp. 444–449 (2013)
10. Krusienski, D.J., Sellers, E.W., McFarland, D.J., Vaughan, T.M., Wolpaw, J.R.: Toward
enhanced P300 speller performance. J. Neurosci. Methods 167, 15 (2008)
11. Choi, K.: Control of a vehicle with EEG signals in real-time and system evaluation. Eur.
J. Appl. Physiol. 112(2), 755–766 (2011)
12. Royer, A.S., Doud, A.J., Rose, M.L., He, B.: EEG control of a virtual helicopter in 3-
dimensional space using intelligent control strategies. IEEE Trans. Neural Syst. Rehabil.
Eng. 18(6), 581–589 (2010)
13. Shi, T., Wang, H., Zhang, C.: Brain Computer Interface system based on indoor semi-
autonomous navigation and motor imagery for unmanned aerial vehicle control. Expert Syst.
Appl. 42(9), 4196–4206 (2015)
14. Cudo, A., Zabielska, E., Bałaj, B.: Introduction to the problem of brain-computer interfaces.
KUL (2011)
15. Lin, Z., Zhang, C., Wu, W., Gao, X.: Frequency recognition based on canonical correlation
analysis for SSVEP-based BCIs. IEEE Trans. Biomed. Eng. 53(12), 2610–2614 (2006)
16. Zhu, D., Bieger, J., Molina, G.G., Aarts, R.M.: A survey of stimulation methods used in
SSVEP-based BCIs. Intell. Neurosci. 2010 (2010)
17. http://emotiv.com/
18. http://openvibe.inria.fr/
An Innovative Approach of Industrial Robot
Programming Using Virtual Reality
for the Design of Production Systems Layout
Radovan Holubek(&), Roman Ružarovský,

and Daynier R. D. Sobrino
Institute of Production Technologies,

Faculty of Materials Science and Technology in Trnava,
Slovak University of Technology in Bratislava, Jána Bottu Street, 25,
917 24 Trnava, Slovak Republic
radovan.holubek@stuba.sk
Abstract. Current trends in the design of manufacturing systems are aimed at

shortening the pre-production phase. Designing CAD (Computer Aided Design)
data and creating simulation scenarios are currently strategic solutions for
designing different variants of manufacturing systems. The research intention is
to point out the possibilities of hardware connectivity and software tools of the
present. The primary intent of this article is the capabilities of industrial robot
programming by utilizing the interaction of several robot online method, OLP
(Offline Robot Programming) and VR OLP computing methods with regard to
the applicability of the proposed outputs to the real environment. As a sup-
porting simulation tool ABB RobotStudio was used. As imaging hardware tools
were used glasses designed to be displayed by the immersive VR environment
HTC Vive. The article presents methods and experiments on a case study where
the interaction between the immersive environment of VR and the area of
classical OLP programming was studied. The immediate interaction of the
classic OLP method of industrial robot programming in the PC desktop software
environment and the immersive VR environment helps to shorten the time
needed to create industrial robot programming. It is precisely such an interactive
link between OLP programming and real-time imaging of an industrial robot in
a VR environment is one of the Industry 4.0 strategy implementation area.
Keywords: Offline robot programming CAD model Robotics simulation

Virtual reality ABB RobotStudio HTC vive
1 Introduction
The Computer Aided Design (CAD) is an indispensable design phase for each pro-
duction system. Using various CAD data and subsequent integration into the simulation
environment often causes a number of different issues in 3D rendering and compati-
bility. Therefore, it is necessary to use various CAD converters that allow migration

https://doi.org/10.1007/978-3-030-18715-6_19
224 R. Holubek et al.
between different CAD formats. The precisely created CAD design is a successful
prerequisite for the subsequent import of the CAD model into the simulation envi-
ronment and then into the immersive Virtual Reality (VR) environment. Therefore, it is
necessary to follow the process of creating used CAD data and integrating this data into
a supportive simulation environment.
Nowadays, in practical industrial applications, there are two main possibilities of
robotic programming methods, which are, online programming (including lead-through
and walk-through) and Offline Programming (OLP) [1]. In general, robot programming
methods are normally classified to:
On-line programming methods, where a robot is used to generate the program,
Off-line programming methods OLP, used where there is needn’t to access the robot
in order to develop the control program. Well at least, no need until the final test of the
program. It is basically consisting of writing a program using a text-based robot pro-
gramming language like ‘Siemens Process Simulate’ or ‘ABB RobotStudio. Figure 1
shows the main standard methods used in the robot programming [2].
Fig. 1. Main standard methods of the industrial robot programming.
Offline robot programming (OLP) is more complex than online programming as the
programming method must have to 3D robot targets but also must have to trajectory of
robot motion and optimize the sequence of the path process generate [3]. The key steps
of OLP are shown at Fig. 2.
An Innovative Approach of Industrial Robot Programming 225
Fig. 2. The most important key steps OLP programming [3].
First of all, in order to achieve this integration, it is necessary the develop a

procedure for the Virtual Prototyping that foresees, in particular, an optimization of the
management of data exchange protocols between 3D parametric CAD systems and
visualization environments using meshed models [4–6].
In the remainder of the period, in addition to the concept of virtual reality of the VR,
other areas have been discovered and added, which are called AR (augmented reality)
wider reality, expanded virtuality AV, and the concept of mixed reality MR. These
terms, as one of the first ones, have been described in their publication in 1994 [7].
These concepts have been crafted over time and advanced through advanced
technology, hardware, software, and computer upgrades to existing support features
such as “headset” VR glasses with indelible hardware - software-based computer
support, smartphones and tablets supporting various advanced reality applications AR,
as well as the mixed reality of the two described areas [8, 9].
Research into industrial robot programming via virtual reality is focused on the use
of classical industrial robot programming methods with the capabilities of linking new
support technology tools to the present [10–12], also used Oculus Rift Virtual Reality
technology [13, 14] with software Unity [15]. Tahriri and colleagues develop the
optimization model for the testing, in which the data are extracted from the VR system
[16]. The results show an increase in production rate and a decrease in cycle time when
the make-span is minimized. The virtual reality robotic teaching system (VRRTS) [17]
offers several benefits to users, and will therefore surpass complex and time-intensive
conventional robot programming methods. Authors [18–20] are concluded that the
virtual reality based solution approach can be implemented in an industrial robotic
work cell.
2 Objectives
As we know from previous research, the issue of virtual reality in engineering is

gradually expanding and introducing itself into the design of manufacturing and robotic
systems. This is related to the issue of digitization within Industry 4.0, where advanced
and virtual reality is one of Industry 4.0’s technology pillars. Simplifying traditional
approaches to robot programming has become one of the most important robotic
research aims. Programming robots using traditional methods is often difficult for
inexperienced and unqualified users. A promising solution for automatic transfer of
knowledge from a human to a robot is the VR environment. The research intention is to
point out the possibilities of HW connectivity and SW tools of the present. The aim of
this article is the possibility of programming of industrial robots using the interaction of
different methods of programming of industrial robots Online, OLP and VR OLP with
regard to the direct applicability of the proposed outputs in real environment. The aim
of such research is planned control program for industrial robot ABB IRB 120 labo-
ratory UVTE MTF STU through VR support [21]. The aim is to define the pro-
gramming methods using virtual reality, plan Path and targets easily, such as play
games designed for VR environment. Nevertheless, the Path and targets must be exact
defined, as is the case with the traditionally used methods of Online and OLP industrial
robot programming.
3 Methods
The experiments will use ABB RobotStudio (RS) software that includes a support
module for viewing created studies in a VR environment. This means that the simu-
lation environment creates a copy of the image of the robotic workstation displayed on
the headset HTC VIVE [22]. In this case, the user is defined as a VR programmer who
has been created in the Steam support for HTC Vive software, and is tracked at all
times throughout the process by movement tracking sensors. The virtual environment
used in the experiments includes a working plane, a set of geometric objects forming a
robotic manipulation workplace with the ABB IRB 120 industrial robot model with the
necessary peripherals. After multiple experiments, it was found that this programming
method allows mutual three-way interaction between the OLP, the RAPID program
and the VR environment. It also allows real-time synchronization, but after a certain
time response. A new approach to visualization a simulation study in a VR environ-
ment to program the trajectories of an industrial robot is to instantly interact during the
creation of individual targets and most of the commands generated in the two visu-
alization environments. However, this interaction requires two users, one as a virtual
reality programmer and the other as a command user through a RAPID program or
OLP software.
Fig. 3. Three-way interactions between OLP, ONLINE and VR environment.
Figure 3 shows a three-way interaction environment between the standard OLP

environment used by OLP programming, the RAPID program via Virtual Flex-pendant
ABB, and immersive environment VR displayed through the headset HTC Vive.
Creating the possibility of using targets consists of instantaneous acts performed in
OLP desktop version, and making changes and creating a new command in immersive
environment in software RS through headset HTC Vive. Haptic hand controllers allow
you to generate the following commands in OLP desktop version RS:
Teleport: is used to visualize different views of the currently displayed simulation

study. Allows you to display complex and commonly invisible views compared to the
PC monitor display. Using Walk Mode, VR headset HTC VIVE lets you capture
motion by viewing sensors located in the corners of the room and visualize it as if it
were to use the Real-Time Online Programming method. This feature is mainly used to
display complex manufacturing systems with the integration of a multitude of com-
prehensive CAD data.
Simulation Control: used to validate the generated simulation as a startup simulation
using the buttons PLAY, PAUSE, STOP. The main purpose of the simulation is to
verify the robot’s reach, to verify with CAD support, and to clear collisions as well as
tune targets and paths.
Interact: Jog is used to guide the robot to the desired position by hand haptic con-
troller by clicking on the active instrument and easily guided into position. With this
function, the robot ranges for the selected application and the suitability of the selected
robot type can be quickly verified in the design of various production systems. Such a
jogging option is unique in the immersive VR environment, as with a robot of a higher
class, such a function is unrealistic and is only possible to use the robot online using the
real flex-pendant method. This represents a very lengthy ride of the real robot into the
required position, not to mention the need to have a real industrial robot available.
Real-time engine unlocking is only possible with small industrial robots (IRB 120 etc.).
For higher class robots it is no possible from a safety point of view. Undo the function
is to go back a step back in the VR environment. Of course, you can also edit the
desktop version of your PC. Redo the function is used to move one step forward in the
VR environment.
Programming: The Path function is used to edit created targets in the following
support functions: Robot Follow: based on the target selection using the haptic hand
controller the active tool robot moves to the desired target; Re-orient using this
function immediate reach response to the robot’s reach to the exact target position. If
the robot does not know how to jump to reoriented target, the show a mistake message;
Trans: immediate change of position of selected target; Fine – Tune: the possibility of
incremental debugging target. Function Teach target can be used in conjunction with
the function Jogging robot. The subsequent creation of individual targets by supporting
function ADD. Generate Paths with the type of Move (Move L, Move J, Move C);
Record: interactive targets and Path generation; Speed: You can change speed even
during running simulation on any Move instruction with interact change in Rapid task;
Zone: Ability to change the zone; Conf.: It is possible to configure rotation of indi-
vidual robot axes with respect to the target the robot has to jump in order to configure
the values on the individual axes of the Rapid task. These individual functions have
been subjected to experiments on a robot model designed for the manufacturing
system.
4 Application
In the first time, we conducted a series of experiments on simulation of component

layout in CAD design with VR support. VR enables us to create a layout with a realistic
3D view. On Fig. 4 is a way to create a workspace layout with VR support.
Fig. 4. Layout design in the Virtual Reality application through the ABB RobotStudio.
The experiments were carried out by performing a “Grab” layout component layout
on the VR programmer and the OLP user importing the CAD models into the layout
directly through the OLP software, where it can perform the import/delete functions of
the individual components, and the VR programmer only safeguards the layout, which
accelerates the creation of the layout itself through the VR environment.
Subsequently, we will discuss how the VR is used to create a Target Path. The,
namely the standard method teach and jog and a novel ABB RobotStudio software
allows you to use two methods for virtual interactive programming within the defined
functions record method, Fig. 5.
Fig. 5. Virtual programming interaction methods in the ABB RobotStudio.
The standard feature is very similar to the online robot guidance method. Instead of
using the joystick via the Flexpendant, the “Jog” function is used, where the Haptic
Hand Controller grabbed the current tool and moved to the required locations. The TCP
coordinate system was automatically moved to the OLP simulation model and gen-
erated “Targets” using the “Teach” function, which included values in a 3D rectangular
coordinate system, axis orientation, default speed, zone, tool, e.g. Adjusting these
parameters will be discussed in the next chapter.
An interesting feature in VR is “Record” function, which, unlike the online and
OLP method, allows automatic generation of Path and Targets through virtual robot
motion control. When a tool is operated via the Haptic Hand Controller, hand move-
ment monitors the movement of the robot and also records individual Paths and Targets
with defined speeds, zones, and so on. These values are automatically written to OLP
and RAPID programs, as can be seen in Fig. 5 top right.
5 Results
A very important phase of design of the automated systems is the debugging of the
control program.
Fig. 6. Debugging process of the created control program through the Virtual Reality.
This phase is important during the creation of the control program for industrial
robots, Fig. 6. This phase is divided into two parts, where the paths, targets, speeds,
zones, and all other important parameters are gradually refined to ensure technology.
This is important because the industrial robot is deployed only to a specific technology
application. On the other hand, the application except quality indicators meet the
Quantification in terms of production time and the number of units.
Virtual reality allows the current version of ABB RobotStudio enter and edit
commands such as create Path, defining of the type of Move, change the Zone, Speed,
Configuration of the Axes and Change the location of the Targets.
Fig. 7. The examples of the Commands form the Virtual Reality.
Other commands it is possible to apply to the control system through the command
user via the OLP software. In our experiment task we create the commands like Wait
Time. In our experiments we found that commands are automatically updated in virtual
reality and in the RAPID program. RAPID commands were also updated in OLP and
also in virtual reality. Thus, virtual reality behaved as if it were a live image of real-time
robot movement. On the Fig. 7 are shown some detail explained examples of the
commands, which were entered through a copy of the robotic system in virtual reality.
6 Discussion
The robotics production system has been validated and tested using a case study, detail
explained in the previous chapter. Several settings and assumptions have been made
during the validation process. Based on these, several conclusions can be drawn from
the investigation. Creating points is intuitive by guiding the robot as in a real online
environment, but the positions are inaccurate. This should be tuned, but online. What
happens anyway standard in classic programming of industrial robots. A very useful
tool is “RECORD”, which allows you to create automatic trajectories with set zones,
speeds, configurations, and motion types. Recognizing errors in virtual reality raises
several issues that do not occur in the real environment. Objects may be closed due to
limited view on the scene, feedback is usually obstructed, and manipulation is usually
not subject to physical law. The created robot path is then rewritten into the real robot
and it’s was validated; the relevant code is derived automatically within the VR, Fig. 8.
The method was applied to a 6-axis industrial robot, the program was launched on a
real robotic cell, and it has been proven that user-friendly VR interfaces derived from
the game world can provide an effective paradigm of robot programming.
Fig. 8. The real industrial robot cell with RAPID program developed through the Virtual
Reality.
7 Conclusion
Synergy between robotics and virtual reality is expected in the coming years. Tradi-
tional programming methods will certainly be alive, OLP simulation will take the lead,
but will be supported by virtual reality tools for better imaging in an immersive
environment. Online programming and real commissioning but can be gradually
shortened, saving money and time. Robot programming in virtual gaming environ-
ments is basically offline programming but combines intuitive interface for robot robots
as in online programming but with the ability to far exceed conventional imagination.
In addition, they appeal to young engineers and technicians for their knowledge of
virtual reality technology. In the future, we would like to deal with the new testing,
creating a general methodology of design through VR, testing inaccuracies and perform
various measurements. Virtual reality (VR) is an example of an offline method aimed at
increasing the intuitiveness of the programming task for the humans in a safe
environment.
Acknowledgments. The research work reported here was made possible by the project KEGA-
021STU-4/2018. Development of a laboratory for the design and maintenance of production
systems supported by the use of Virtual Reality.
References
1. Ang Jr., M.H., Wei, L., Yong, L.S.: An industrial application of control of dynamic behavior
of robots - a walk-through programmed welding robot. In: Proceedings of the 2000 IEEE
International Conference on Robotics Automation, San Francisco, CA, vol. 8 (2000)
2. Occupational Safety and Health Administration, Osha Technical Manual (OTM), Section IV:
Chapter 4, Industrial robots and robot system safety (2017). https://www.osha.gov/dts/osta/
otm/otm_iv/otm_iv_4.html
3. Pan, Z., Polden, J., Larkin, N., Van Duin, S., Norrish, J.: Recent progress in programming
methods for industrial robots. Robot. Comput.-Integr. Manuf. 28(2), 87–94 (2012)
4. Caputo F., Di Gironimo G., Marzano A.: A structured approach to simulate manufacturing
systems in virtual environment. In: XVIII Congreso International de Ingegneria Grafica,
Barcelona, Spain (2006)
5. Dornberger, R., Inglese, T., Korkut, S., Zhong, V.J.: Business Information Systems and
Technology 4.0 - New Trends in the Age of Digital Change, Studies in Systems, Decision
and Control. Springer, Heidelberg (2018)
6. Krot, K., Kutia, V.: Intuitive methods of industrial robot programming in advanced
manufacturing systems. In: Advances in Intelligent Systems and Computing, vol. 835.
Springer (2019)
7. Milgram, P., Takemura, H., Utsumi, A., Kishino, F.: Augmented reality: a class of displays
on the reality-virtuality continuum. In: Telemanipulator and Telepresence Technologies, vol.
2351 (1994)
8. Pan, Z., Cheok, A.D., Yang, H., Zhu, J., Shi, J.: Virtual reality and mixed reality for virtual
learning environments. Comput. Graph. 30, 20–28 (2006)
9. Michas, S., Matsas, E., Vosniakos, G.C.: Interactive programming of industrial robots for
edge tracing using a virtual reality gaming environment. Int. J. Mechatron. Manuf. Syst. 10
(3), 237–259 (2017)
10. Chong, J.W.S., Nee, A.Y.C., Ong, S.K., Youcef-Toumi, K.: Robot programming using
augmented reality: an interactive method for planning collision-free paths. Int. J. Robot.
Comput.-Integr. Manuf. 25(3), 689–701 (2009)
11. Jen, Y.H., Taha, Z., Vui, L.J.: VR-based robot programming and simulation system for an
industrial robot. Int. J. Ind. Eng. 15(3), 314–322 (2008)
12. Crespo, R., Garcia, R., Quiroz, S.: Virtual reality application for simulation and off-line
programming of the mitsubishi movemaster RV-M1 robot integrated with the oculus rift to
improve students training. Proc. Comput. Sci. 75, 107–112 (2015)
13. Roßmann, J., Schlette, C., Wantia, N.: Virtual reality providing ground truth for machine
learning and programming by demonstration. In: ASME International Design in Engineering
Technical Conference Computers and Information in Engineering (IDETC/CIE), pp. 1501–
1508 (2012)
14. Mortimer, M., Horan, B., Joordens, M.: Kinect with ROS, interact with oculus: towards
dynamic user interfaces for robotic teleoperation. In: 11th System of Systems Engineering
Conference (SoSE), pp. 1–6 (2016)
15. Crespo, R., García R., Quiroz S.: Virtual reality simulator for robotics learning. In:
International Conference on Interactive Collaborative and Blended Learning (ICBL),
Mexico City, pp. 61–65 (2015)
16. Tahriri, F., Mousavi, M., Yap, H.J., Siti Zawiah, M.D., Taha, Z.: Optimizing the robot arm
movement time using virtual reality robotic teaching system. Int. J. Simul. Model. 14(1), 23–
38 (2015)
17. Matsas, E., Vosniakos, G.-C.: Design of a virtual reality training system for human–robot
collaboration in manufacturing tasks. Int. J. Interact. Design Manuf. (IJIDeM) 11(2), 139–
153 (2017)
18. Yap, H.J., Taha, Z., Md Dawal, S.Z., Chang, S.W.: Virtual reality based support system for
layout planning and programming of an industrial robotic work cell. PLOS One 9(10),
e109692 (2014)
19. Gogouvitis, X.V., Vosniakos, G.-C.: Construction of a virtual reality environment for robotic
manufacturing cells. Int. J. Comput. Appl. Technol. 51(3), 173–184 (2015)
20. Wang, H., Mecham, R., Zhang, B.: A method targeting repair in space: tele-operating a
collaborative robot with virtual reality. In: IEEE 7th Annual International Conference on
CYBER Technology in Automation, Control, and Intelligent Systems (CYBER), Honolulu,
HI, pp. 1068–1071 (2017)
21. Holubek, R., Ružarovský, R., Delgado Sobrino, D.: Using virtual reality as a support tool for
the offline robot programming. Res. Pap. Fac. Materials Sci. Technol. Slovak Univ. Technol.
26(42), 85–91 (2018)
22. Holubek, R., Sobrino, D.R.D., Košťál, P., Ružarovský, R., Velíšek, K.: Using virtual reality
tools to support simulations of manufacturing instances in process simulate: the case of an
iCIM 3000 system. In: MATEC Web of Conferences, vol. 137 (2017)
Localization of Passive RFID Tags by Small
Cartesian Robot
Tomasz Kapłon(&) and Andrzej Milecki
Poznan University of Technology, Pl. M. Sklodowskiej-Curie 5,

60-965 Poznan, Poland
tomasz.z.kaplon@doctorate.put.poznan.pl
Abstract. In this paper simple localization method is proposed which uses

radio-frequency identification (RFID) system working with low frequencies.
The position of RFID tag is obtained by the recognition system installed on a
small cartesian robot. In the robot head a RFID reader is installed. It was
assumed that the used components for localization should be inexpensive. In the
localization the inversion of the typical method based on proximity is applied.
The difference is that the position of moving reader is not examined by tags, but
position of tags is recognized by moving reader. The method was able to obtain
precision in a range of a few millimeters. Two variants of collecting detection
points and two variants of calculating position were examined. It was examined
how the distance between the reader and the tag influences on position error, and
what is the influence of head velocity on the detection accuracy.
Keywords: RFID Passive tags Indoor localization
1 Introduction
Today problem of indoor localization of different objects is still a vital problem in

many automated devices. Although, outdoor localization, thanks to the systems like
Global Positioning Systems (GPS), is successfully implemented especially for big
objects like cars. There is still lack of good solution for estimating precise location of
elements with accuracy of a few millimeters. In this paper the possibilities of solving
this issue by using RFID system are proposed.
Since the beginning of the XXI century different methods of localization of RFID
tags were developed [1]. Most of them can be categorized in four groups. These groups
can be described as: based on distance estimation, scene analysis, constraint based
approach, proximity [2].
Methods from the first group are based on trilateration and triangulation. In order to
use those methods there is necessary to know distance and angles between the reader
and tag. In order to estimate the distance usually such methods like Time of Arrival
(TOA), Time Difference of Arrival (TDOA) or Received Signal Strength Indicator
(RSSI) [2–4]. Angles are usually calculated using the information about signals phase
shifts [2]. The drawback of methods based on TOA, TDOA and phase shifting is the
fact that precise time measuring and synchronization are for them crucial. Because of
that there is a need of special hardware for that methods [4].

https://doi.org/10.1007/978-3-030-18715-6_20
Localization of Passive RFID Tags by Small Cartesian Robot 237
The group of methods described as scene analysis is also known as fingerprinting.

LANDMARC is an example of these methods [5, 6]. In this approach usually special
maps of RSSI in defined area are prepared, and then during online phase, they are used
to estimate position of tag placed in this area [2]. There are a few subgroups with a
different approach of proceeding second phase. In K-Nearest method, in the first step,
the value of RSSI is used to select k nearest reference tags in relation to a tag with an
unknown position. then the searched position is determined as the weighted average of
the positions of the selected reference tags. The probabilistic method in the second
phase assumes the existence of n possible locations and one vector of signal strength.
The specific location is determined by determining the most likely right point. In this
purpose can be used, for example, Bayesian localization [7]. The newest method is
datamining approach. In that method RSSI collected in the first phase is classified and
used in offline learning. In the second phase neural network is used to find matching
patterns for incoming data [4].
The third popular approach is described as constrain-based. In this case a few
readers are used and information about the position of tag is obtained as the combi-
nation of information from all readers, whether the tag is in the operational area of the
reader or not. The method, that belongs to that group, called L-VIRT was presented in
[8]. The group of the simplest methods of localization, that uses RFID systems, is
described as based on proximity. Usually in these methods determined area is paved
with tags with defined positions. The moving object is equipped with reader or only
with an antenna. When a tag is detected by the antenna, its position is assumed as
position of moving object. In case of detection of two or more tags additional algo-
rithms are used in order to establish the position [6, 9]. This method works in similar
way if the antennas are densely arranged and the tag is tracked [2]. The advantage of
this method compared to the RSSI-based ones is that the RSSI values may be more
unstable compared to the probability of receiving a tag’s response [10].
For localization purpose both active and passive tags are used. Active tags have
own sources of energy, for example batteries. Thanks to this they can work indepen-
dently from reader and give stronger signal. However these tags are more expensive.
Passive tags need energy delivered by the reader in the form of radio waves. That limits
their possibilities, but makes them also cheaper and smaller. Because of that, today
systems which use passive tags are being developed. On the market additional group of
tags is specified, which is called as semi-passive. They are equipped in own low-power
source of energy, which is used only to maintain storage of data in memory. The whole
process of communication of these tags must be powered by the reader [11].
Another approach to the problem of using RFID to localize object was presented in
the work [12]. Authors used only one reader which follow determined road in the form
of parallel lines and makes consecutively read operations. The only collected data for
every read operation was the number of the scan line, ordinal number of read operation
in line and identification data of detected tags. Thanks to that it was possible to
determine the position of RFID tag in the matrix of rows and columns. However, this
method needs a few assumptions. The number of columns and rows must be known,
and number of tags in every row must be equal [12]. These assumptions are the main
limitations of using the method.
238 T. Kapłon and A. Milecki
The main drawback of existing methods of tags position calculation is their poor
precision. Some of them are able only to determine that the tag is present in the area of
antenna range. Another problem of known methods is their small ability to maintain
their performance in changing environment. In order to overcome this problem,
investigations of using of artificial neural networks are undertaken, as described in [13].
Today usually Ultra High Frequency (UHF) or other advanced RFID systems are used
for tag position recognition. In described in papers solutions low frequency (LF) RFID
systems are not described, therefore in this paper the application of LF RFID tags is
proposed.
The aim of this work is to develop method that uses inexpensive LF RFID components
in localization purpose. It was decided to use one of the simplest accessible on the
market LF RFID system, called Unique that work with 125 kHz frequency and have no
anti-collision mechanism. That means that the reader is able to read only one tag in its
reading zone. For the chosen system data frame has 64 bits of length and the time
necessary for reading one bit equals 512 ls. In the research the cartesian robot, with
stepper motors as drives is used. It moves the RFID reader above the area in which tags
are located. A road of the reader is determined and may be freely programmed by the
user. Thanks to that, the reader position is known in every moment. Due to this fact, if
the reader enters area of interacting with a tag and reads its code, coordinates of this
point are known. The nominal single step of the robot is equal around 0.18 mm. The
values of the position coordinates obtained in this way are influenced by a few errors.
One of them is caused by the fact that there is a delay between entering the zone of
interacting with tag by the reader, and reading out an identification code. This is
because the reader needs time to deliver energy to the tag and then needs additional
time for receiving complete code from tag. During a time of this delay, the reader
travels relatively long distance, especially when data transmission speed is low and
speed of movement of the reader is high. Another factor which can influence on the
system accuracy is that the antennas surfaces of the tag coil and the reader can be not
perfectly parallel. Because of that, depending on the direction of movement, the tag can
be detected as being closer or further i.e. sooner or later. Moreover in most cases, the
area of interacting of the tag and the reader do not cover the boundaries of tag. It is a
result of propagation of radio waves in time. Because of this, one detection point does
not give accurate information about position of the tag. Therefore, in the next step of
the investigations it was examined how the collection of that points coordinates
obtained from different sides of tag may improve the accuracy of the results.
In the first step of the described here investigations, the influence of the speed
movement of the reader, on tags detection points distribution was examined. In these
investigations two types of tags were used, i.e. with diameters of 25 mm and 15 mm.
As it is visible in Figs. 1 and 2, when velocity of reader is relatively small, the detection
points form deformed circle with offset regarding to the center of tag. In order to
compare the obtained results, the calculated average values of the coordinates for every
reader’s speed are presented in the Table 1.
speed: 0,095 m/s speed: 0,114 m/s

Altitude :10 mm
Road of reader
speed: 0,26 m/s
Fig. 1. Distribution of detection points for 25 mm of diameter tag
It can be noticed that the error for average points becomes smaller when speed of
reader is bigger. It is result of the fact that detection points are concentrated closer to
the center. Moreover, although single detection point can be close to the center of tag,
the dispersion of possible errors is very high.
In the second step of the investigations, a correlation between the error of calculated
position and the number of detection points was examined. It was examined how big is
error between nominal position of center of tag and average point in relation to number
of points used to calculate average. The examinations ware conducted for the speed of
reader movement of 0,19 m/s. The altitude between the head and the RFID tags with
diameter of 25 mm, was equal to10 mm.
It was also assumed that value of position error can be estimated with modified
formula of error for average (1):

Altitude :10 mm
Road of reader
speed: 0,26 m/s
Fig. 2. Distribution of detection points for 16 mm of diameter tag
Dðd; v; hÞ
D¼ pffiffiffi ð1Þ
2 n
where: Δ-position error, n-number of detection points, D-ostensible diameter, v-

velocity of RFID reader, h-altitude of the reader over tag, d-diameter of the tag.
As it can be seen in Fig. 3, the value of position error is correlated with the number
of points used to obtain average coordinates of the point i.e. tag. It is visible, that the
use of 3 or 5 points is sufficient and higher number of points does not give
improvement of the results. Also, when the number of points is smaller than 3, the
Table 1. Results of examination of distribution of detection points

Error for detection points Error for average point
Speed [m/s] Standard deviation Variance Min [mm] Max [mm] Δ [mm]
Diameter 25 mm
0,095 6,7 44,8 4 20 8,9
0,114 6,7 45,2 4 21 8,5
0,14 6,3 39,5 4 20 8,8
0,19 5,1 26,2 4 17 6,6
0,26 3,9 15,3 5 16 5,4
Diameter 16 mm
0,095 3,8 14,3 5 14 4,9
0,114 5,0 24,8 4 16 7,1
0,14 4,3 18,5 4 14 6,7
0,19 3,9 15,0 4 13 5,8
0,26 4,4 19,6 1 14 2,3
Fig. 3. Relation between number of detection points and position error
obtained position error increases. It can also be seen that maximal difference between
experimental results and obtained from model is around 1 mm. Taking into the
obtained and described above results, the concept of the road for the reader was
developed. It was decided that the reader will scan the area in which the tags are
located, moving along parallel lines. In the case of recognition the tag by the head, the
distance between those parallel lines will be reduced and the head will retract and start
the recognition procedure again, following a loop around the tag location. It was
assumed that such procedure will guarantee that there will be at least three detection
points and they will be taken from different sides of the tag. After closing the loop, the
position of tag is calculated. In Fig. 4 the example of the reader road is presented.
Fig. 4. Road of the reader following special loop. 1-reader, 2-tag
The disadvantage of this method is that in one direction it does not take advantage
with full resolution of stepper motor used as a drive.
Because of that second scheme of the road of reader was proposed. In this approach
the aim is to collect pairs of detection points that lie on two perpendicular lines parallel
to axis x and y. As the result reader after meeting tag follow a road similar to cross. The
road is presented at Fig. 5.
Fig. 5. Scheme of the second variant of the road of the reader. Road similar to cross 1-reader,
2-tag
Taking under consideration, the fact that distribution of detection points can form
shape similar to circle, the second method of calculating of position of tag’s center was
proposed. Having three points situated at circumference, it is possible to build a tri-
angle. The center of circle can be calculated as a point of intersection of the perpen-
dicular bisectors of sides of the triangle. Having more than three points it is possible to
build more triangles and obtain more possible answers, according to the well known
equitation for number of possible combinations (2):

n n!
i¼ ¼ ð2Þ
k k!ðn kÞ!
where: k = 3, n – number of points, i – number of possible centers of circle.

In the method that was proposed, it was decided to obtain final center of circle as
the average value of the series of possible locations.
3 Experiment Results
In the investigation it was examined how the altitude of the reader location over tag
influences on position error. In the first attempt, the RFID tag was placed in the location
with known coordinates, and then it was detected by developed system. In the next step
the influence of the velocity of the reader on calculated tag’s position error was
Fig. 6. Results of investigation of the influence of altitude on position error (velocity of the
reader 0,19 m/s).
Fig. 7. Results of investigation of influence of the velocity of the reader on position error
(altitude of the reader 10 mm).
Fig. 8. Influence of altitude on position error for points calculated as average and as center of
circle, velocity of the reader 0,19 m/s
examined. The center of the tag was calculated as average value of coordinates of
detection points. Position error was calculated as a distance between calculated average
point and nominal center of the tag. The examination was done for tags with diameter
25 mm and 16 mm.
Fig. 9. Influence of the speed of reader on the position error for points calculated as average and
as center of circle (altitude of the reader 10 mm).
For comparison of two methods of calculating position of tag, data that were
obtained from first method of collecting detection points were used in method of
calculating center of the tag that is based on fitting circle to the set of points.
Fig. 10. Comparison between influence of the speed of the reader on position error for points
collected with the reader following the road in the form of loop (Tag 25) and reader following the
road like cross (Tag 25’)
For comparison of two methods of collecting detection points, they were examined
for tag with the diameter of 25 mm, and at reader’s altitude of 10 mm. The center of
the tag was calculated as average value of the coordinates of detection points (Fig .9).
4 Discussion and Conclusion
4.1 Discussion
As it can be seen at Figs. 8 and 9, method based on finding the center of circle with
using three points can give poor results. During the examination in some cases only
three points where collected and because of that only one possible center could be
found. Moreover, when additionally distribution of detection points is not similar to the
circle due to the high velocity of the reader, huge errors occurred. For this reason, this
method is less reliable than the method of finding the tag center as an average of the
coordinates of the detection points.
It can be observed that, generally together with the growth of the altitude of the
reader position above the tag, also the position error grows. However, the best results
were achieved not for the minimum value of the altitude, but for 10 mm. It seems that it
is optimal distance for that configuration of tag and reader with antenna. The area of
interaction is big enough that reader has enough time to deliver energy to the tag and
receive its code, near the boundary of the tag.
As it is shown at Figs. 6 and 7 better accuracy was obtained for tag with 25 mm
diameter. However increase in the velocity caused bigger improvement for tag with
16 mm diameter. Finally the smallest position error for 25 mm tag was around 2 mm
and for 16 mm tag it was around 3 mm. Those values of position errors are much
smaller than values that can be achieved by RFID localization systems that only use
fixed readers.
4.2 Conclusion
In this paper a simple method of localization that uses RFID tags was proposed. Thanks
to the synergic work of mechanical system and RFID reader it was possible to obtain
accuracy in the order of a few millimeters. In the best cases the error was only around
2 mm. This result is better than results achieved by systems using only RFID readers.
Moreover the presented system uses only chip RFID elements that can reduce costs of
implementation. The main limitation of the method is fact that the position measure-
ment and calculation needs time and therefore the position of tag can’t be delivered
online.
The best results were achieved for method based on the movement of the reader,
following special loop and calculation of the center of the tag as average value of
coordinates of detection points. It was found out, that good results can be achieved for
that method even if only three points were collected. It can be state that method of
finding the center of the tag based on the fitting circle is not reliable. The main reason
of this, is that for high speeds of reader’s movement, detected points do not form a
circle. To improve the detection accuracy more lines of scanning should be used or
velocity of reader should be small, which will increase time of examination of one
point.
One of the main obstacles for accurate localization of the tag by fast moving reader
is that in dynamic conditions, a few detection points can be missed. Generally for
reader moving this same road, together with growth of speed less and less points are
detected. That means that good method of localization should be resistant to possibility
of missing one or two predicted detection points. In order to improve the results of tags
detection some changes in robot construction should be considered. Replacing stepping
motors with motors that work continuously like BLDC and adding position sensors can
be one of them.
References
1. Valero, E., Adán, A.: Integration of RFID with other technologies in construction.
Measurement 94, 614–620 (2016)
2. Singh, S., Aggarwal, A.: Survey on localization techniques of RFID for IOT. Int. J. Comput.
Appl. (0975 – 8887) 137(12), 23–27 (2016)
3. Dmitry, A.: Savochkin, simple approach for passive RFID-based trilateration without offline
training stage. In: 2014 IEEE RFID Technology and Applications Conference (RFID-TA)
(2014)
4. Bilodeau, J.-S., Bouzouane, A., Bouchard, B., Gaboury, S.: An experimental comparative
study of RSSI-based positioning algorithms for passive RFID localization in smart
environments. J. Ambient Intell. Humaniz. Comput. 9(5), 1327–1343 (2018)
5. Ni, L., Liu, Y., Lau, Y., Patil, A.: LANDMARC: indoor location sensing using active RFID.
Wirel. Netw. 10(6), 701–710 (2004)
6. Bouet, M., dos Santos, A.L.: RFID tags: positioning principles and localization techniques.
In: 1st IFIP Wireless Days, WD 2008 (2008)
7. Park, J.-G., Charrow, B., Curtis, D., Battat, J., Minkov, E., Hicks, J., Teller, S., Ledlie, J.:
Growing an organic indoor location system. In: Proceedings of the 8th International
Conference on Mobile Systems, Applications, and Services, 15 June 2010, pp. 271–284
(2010)
8. Bouet, M., Pujolle, G.: L-VIRT: range-free 3-D localization of RFID tags based on
topological constraints. Comput. Commun. 32, 1485–1494 (2009)
9. Yang, P., Wu, W., Moniri, M., Chibelushi, C.C.: Efficient object localization using sparsely
distributed passive RFID tags. IEEE Trans. Ind. Electron. 60(12), 5914–5924 (2013)
10. Savochkin, A.A., Mickhayluck, Y.P., Iskiv, V.M., Schekaturin, A.A., Lukyanchikov, A.V.,
Savochkin, D.A., Levin, E.A.: Passive RFID system for 2D indoor positioning. In: 2014
20th International Conference on Microwaves, Radar and Wireless Communications, pp. 1–
3, June 2014
11. Oguntala, G., Abd-Alhameed, R., Jones, S., Noras, J., Patwary, M., Rodriguez, J.: Indoor
location identification technologies for real-time IoT-based applications: an inclusive survey.
Comput. Sci. Rev. 30, 55–79 (2018)
12. Son, Y., Joung, M., Lee, Y.-W., Kwon, O.-H., Song, H.-J.: Tag localization in a two-
dimensional RFID tag matrix. Future Gener. Comput. Syst. 76, 384–390 (2017)
13. Soltani, M.M., Motamedi, A., Hammad, A.: Enhancing cluster-based RFID tag localization
using artificial neural networks and virtual reference tags. Autom. Constr. 54, 93–105 (2015)
Analysis of Cooperative Industrial Task
Execution by Mobile and Manipulator Robots
Khurshid Aliev and Dario Antonelli(&)
DIGEP, Politecnico di Torino, Corso Duca degli Abruzzi, 24, 10129 Turin, Italy
dario.antonelli@polito.it
Abstract. The emergence of mobile robots as a flexible upgrade of industrial

AGVs and the simultaneous diffusion of collaborative manipulators pose new
problems for the organization of work in industrial plants. The new robots
address work environments characterized by limited automation and unstruc-
tured layouts. Present study is aimed at demonstrating that, using commercially
available technologies, it is possible to assure a fruitful collaborative interaction
among three main actors of the factory of tomorrow: the human operator, the
mobile robot and the manipulator.
Keywords: Human robot collaboration Computer vision

Autonomous robot Industrial assembly
1 Introduction
Autonomous mobile robots and collaborative manipulator robots are two sides of an
emerging Industry 4.0 enabling technology that aims at modifying the work organi-
zation of manufacturing and assembly workcells by allowing the collaborative work of
humans and robots [1]. While conventional industrial robots are complex, expensive,
and require large shop floor space for deployment, collaborative robots (cobots) are
easy to program, inexpensive, and intended to work alongside, interact, and assist
human workers, thereby creating an entry point for introducing automation in SMEs.
The reasons for their diffusion are the increased human safety, the easy deployment and
fast setup, the unprecedented flexibility due to the fast programming and to the col-
laboration with human workers, and, last but not least, to the reduced costs as they
don’t need dedicated and exclusive areas in the shop floor.
While robot manufacturers produce collaborative robot models since long time,
their introduction in industry was made possible only recently. An important milestone
has been the issue of the ISO Technical Standard 15066, defining safety requirements
for collaborative industrial robot systems and the work environment. It was a necessary
integration of the safety standards ISO 10218-1 and ISO 10218-2. In [2] ISO-TS 15066
standard is explained through the help of an experimental application. Nowadays, it is
possible to employ collaborative robots in standard industrial environments with the
guarantee of safe and reliable operativity.
A further step in the revolution of factory automation will be the combination of
autonomous robot with manipulator arm to increase furthermore the range of processes

https://doi.org/10.1007/978-3-030-18715-6_21
Analysis of Cooperative Industrial Task Execution 249
that can be operated and the production flexibility. This new type of industrial robot is
called autonomous or mobile manipulator. The collaboration now involves three actors,
human, robotic arm, mobile robot. Each of them has a different ‘brain’ and interact with
the external environment through a different system of sensors.
There are research projects that investigate different aspects of this topic: system
stability [3], localization [4], navigation [5], energy consumption [6]. The introduction
of these mobile manipulators in factories is still infrequent. In present study, the
application of technologies and methods outside research laboratories has been
investigated.
A commercial manipulator and a mobile robot have been employed. The additional
sensor needed to enable the collaboration between the two robots has been a com-
mercial internet camera. Realistic pick and place tasks have been executed using dif-
ferent parts with different materials and dimensions, under lighting conditions non-
optimized for computer vision, i.e. in a factory-like context. Thus, the feasibility of
autonomous cobots introduction in a realistic shop floor has been tested. As a verifi-
cation of the accuracy, the robot position has been measured with an infrared multi-
camera 3D tracking system as reference system.
Section 2 describes the state of the art on which present experiment relies. Sec-
tion 3 describes the equipment and the case study. Section 4 the methods adopted to
recognize and localize the workpiece, and the application to a task-based programming,
Sect. 5 presents results and discussions. Section 6 contains the conclusions.
2 State of the Art
The section presents existing methods to cope with the problems of workpiece
detection, recognition and localization in the context of industrial applications. Addi-
tionally, some issues concerning robot flexible programming for the execution of
industrial applications are discussed.
The first industrial use of computer vision is to verify if the object is present in the
work area. Background subtraction is probably the most adopted approach [7], fol-
lowed by image segmentation [8] or saliency detectors [9]. Background subtraction
requires to memorize an empty image of the workspace (background) and then to
highlight the part of the image that is changed.
The second problem is the recognition and localization of the workpiece. It does not
imply necessarily the use of vision techniques. Laser scanners are used by most
autonomous mobile robots and 3D range imaging and scanners are increasingly applied
to different fields. [10] presents an updated review of modern 3D data acquisition
techniques. There is another non-visual and inexpensive method to recognize object
type and identity, by mounting on the workpiece an RFID [11]. Vision-based local-
ization methods don’t require the pre-processing of the part, contrarily to RFID, but are
sensitive to the shape of the part, the orientation with respect to the camera, the likely
presence of reflective surfaces and eventually to the room lighting conditions. State of
the art for factory applications are methods that operate a matching with features,
derived by CAD [12] or extracted automatically during a pretraining phase [13], known
as Viola-Jones classifier. Viola-jones classifier is reliable even under non-optimal
250 K. Aliev and D. Antonelli
lighting conditions. Therefore it was selected for present research, using the method
[14] to deal with reflective surfaces.
Concerning the last problem, how to make the robot execute task-based programs
with enough flexibility to be adaptable to variable workpieces in variable positions,
there is poor evidence of specific literature. Task based programming is a programming
method well studied, but seldom in the contest of industrial applications. Thus, recently
several studies allowed to work out all the issues present in this programming tech-
nique. [15] allowed the debug and improvement of the program by assuring an aug-
mented reality feedback. [16] used a probabilistic approach to cope with the inevitable
disturbances present in an industrial assembly positioning. [17] deals with the teaching
phase that in an industrial application should be simple, fast and moreover reliable. In
present experiment the task-based programming has been considerably facilitated by
the adoption of a hierarchical formalization of the assembly operations provided in
[18].
The problem of task assignment to human operators and to robots has been solved
by using the classifier presented by [19] based on C4.5 decision tree that has been
trained on a set of samples taken from existing industrial assembly processes.
3 Case Study and Equipment Description
The laboratory case study was developed after several discussions with automation
managers from robot manufacturing companies and from system integrators. The
equipment employed is made of:
1. Randomly positioned workpiece on the surface of the mobile robot,
2. Mobile Industrial Robot (MIR100) used to transport the workpiece that need to be
transported to the working area,
3. Vision system (Logitech C920 camera) mounted on the end effector of the
Universal Robot that detects the location of random placed workpiece,
4. Manipulator (Universal Robot - UR3) - collaborative robot arm to execute pick and
place tasks.
Industrial like workpieces of different materials (reflective metal and dark opaque
polylactide, PLA) have been selected for our experiment, to test the method on a wide
range of possible industrial applications. Reflective workpiece is a mechanical flange
while the PLA one is a 3D printed bracket. The geometry of the flange has additional
features, a complex shape and cylindrical holes. In Fig. 1 both workpieces are pre-
sented. The metal flange is just a part in a complex assembly task that is executed
collaboratively by the robot manipulator and a human operator. The PLA bracket is just
a benchmark part to test the vision system. The assembly task program is generated
automatically combining a set of elementary assembly operations. The assembled
product is presented in Fig. 2.
The autonomous mobile robot (AMR) is employed to transport workpieces from
two locations: origin and destination. AMR navigates via self-constructed maps that its
software generates from the previous navigations. AMR combines data from integrated
cameras, proximity sensors and laser scanners. It is able to detect obstacles and choose
Fig. 1. Sample workpieces: (a) reflective metal flange and (b) dark plastic bracket.
Fig. 2. Complete assembly sequence for the metal flange of Fig. 1a.
the most efficient route to the target. It is therefore easy and cost-efficient to implement
in a non-structured shop floor, because it can reconfigure the trajectory to avoid the
obstacle, selecting an alternative route. Self-planning trajectory imply that the final
position and orientation of the AMR is set up with some uncertainties, depending by
the approaching direction to the target.
To accomplish the experiment of pick and place tasks for light workpieces, a
collaborative robot arm and a two-finger RG2 gripper have been used. RG2 is a force
sensitive gripper, with gripping force selectable between 3 and 40 N. To detect the
arrival of the AMR to the working area, a proximity sensor is sufficient in factory
conditions. In the experiment it was replaced by a manual-operated button switch
connected to the Digital Input of the manipulator controller to have a better control of
the experiment. The experiment components are presented in Fig. 3.
The trials were conducted in a laboratory, with artificial fluorescent lights typical
for a workshop. Illumination is not uniform nor diffuse and this poses additional
problems to the vision system. The experiment area is a robot cell which hosts robots, a
Fig. 3. Group photo of equipments and devices employed in the experiment.
work bench, a PC running image processing software, a wi-fi network to communicate

with mobile robots, and safety fences, a legacy of the past.
The scenario of the case study is described in the following lines. Before starting
the experiment, human operator run the developed software for vision system to detect
a workpiece on PC and the program on teach pendant of manipulator to execute pick
and place tasks. Afterwards, manipulator and vision system software are in waiting
condition. Human operator indicates the origin and destination position to the AMR
that goes to point A (origin) location to bring the workpiece to the working area.
Human operator loads the workpiece. AMR brings the workpiece to the working area
where manipulator is waiting with the vision System mounted on the arm before the
wrist. Whenever AMR arrives to working area manipulator detects it and run the
Vision System to detect the workpiece. After detecting the object, a deep network
classifies it and a computer vision algorithm calculates the location of the object. The
object type and the coordinates are passed to the manipulator control. It will execute the
pre-programmed task corresponding to the workpiece, using the location of its center of
mass. Distribution of work activities is shown in Fig. 4.
Fig. 4. Activity diagram of the work load distribution among human operator, AMR,
manipulator and vision system.
4 Classification Method and Localization Algorithm
The adopted method can be decomposed in the following phases:

1. line up tool center point (TCP) of the manipulator with the vision frame
2. communication of the reference frame to both the robot and the vision system
3. detect a new object against the background and compute the workpiece position and
center of mass
4. classify the image with reference to known parts to be produced.
4.1 Line up Robot Arm End Effector with Camera Frame

The method used for aligning the vision system frame with the robotic arm is to create
a user defined reference frame. Figure 5 depicts that the reference point of the user
defined plane is aligned to the reference frame of the vision system. The camera frame
is directly mapped to the reference point of the user defined reference frame and the
offset of the TCP relative to the feature frame is calculated using pose_trans() of the UR
script.
4.2 Communication Between Vision and Robot Arm

The aim of this section is to describe the communication between vision system and a
Universal-Robots arm. The description will utilize an application where workpiece is
placed on AMR and workpiece identified by the vision system. The position data and
size of the workpiece is then transferred to the Universal-Robots arm to perform pick
and place task. In order to perform communication between the Vision system and The
Fig. 5. Line up camera frame with user defined feature of robotic arm.
Fig. 6. Process flow of the workpiece recognition, localization on the AMR and pick and place
with the manipulator.
Universal-Robots, TCP/IP protocol is deployed. Vision system interface is written in

python language and identification and localization methods are accessed from
OpenCV libraries. Figure 6 shows the process flow of the case study, the communi-
cation method and script examples are also displayed in the red boxes.
4.3 Detect a New Object Against the Background and Compute

the Workpiece Position and Center of Mass
To develop a software of detection location of the workpiece the background sub-
traction method was selected. The method is very simple: it subtracts background frame
from foreground frame that is captured in Δt time and it applies image threshold
method to binarize the image. The camera is mounted on the UR arm end effector and
calibrated vision system capture frame with user defined plane of UR arm. Thus, the
same coordinated reference frame is adopted by UR arm and by camera, as shown in
Fig. 4.
The implementation of the method is the following: vision system is waiting a
signal from UR arm when UR arm is running, camera captures background image and
when AMR is arrived with workpiece to working area, UR arm controller triggers
vision system to detect the workpiece against foreground image and the location
position of the workpiece on the user defined plane is estimated using conversion pixels
vs. centimeters. In camera frame we estimate the location of the workpiece in 2D
dimension and afterwards we send to UR arm defined frame plane to execute
Fig. 7. Workpiece detection and localization flowchart.

manipulation. The interaction system between Vision and robots is developed in

Python using OpenCV libraries and communication between UR arm and vision
system is TCP/IP protocol. Figure 7 shows the sequence of the vision system program.
5 Qualification of the Experimental Method
The experiment was repeated several times with a 100% percentage of success. The
instant when the manipulator grasp a part is presented in Fig. 8. This was due to the
simplicity of the task and to the lack of disturbances that could happen during industrial
work. To give indications useful for a future implementation in industrial production, a
qualification study of the system was executed by evaluating the error chain in the
experiment. The following elements were subjected to qualification:
1. AMR positioning error
2. Workpiece localization error made by the vision system
3. Classification error of the workpiece.
Fig. 8. The manipulator grasps the flange after having recognized the object and its position.
The AMR positioning repeatability was measured by 50 repetitions of the travel

from origin to destination with varying pathways. The arrival position and orientation
has been measured using Optitrack infrared multi-camera 3D tracking system. To
execute measures, 5 reflective spherical markers have been mounted on the AMR, as
can be seen in Fig. 3. The error distribution is shown in Fig. 9.
Fig. 9. Positioning error distribution relative to the mean value.
During the experiment the average absolute positioning error resulted in X-

axes = ±30 mm, Y-axes = ±55 mm and RX = ±76.6 mm. Max error value 160 mm
and Min. error 0 mm in X-axes. Max. error value, 183 mm and Min. error value, 8 mm
in Y-axes. Rotation error from X-axes was 2.0° and min is 0.0 average absolute
rotational error is 0.7°. The error distribution graph makes evident the presence of a
reduced number of measures with a positioning error far larger of the average. They
can be considered outliers as they could make impossible to execute the pick and place
task. In this case, the AMR should go slightly back and repeat the approaching
movement.
Localization of the workpiece is the second qualification point. It is verified by
determining the associated reference frame. Repeatability of the reference frame
acquisition was measured experimentally. Vision system was placed 1370 mm away
from the origin, inclined 36° with respect to the table surface, N = 50 measurements of
the reference frame were executed with the same relative position of the reference
frame and the tracking system.
The actual standard deviations of the origin location are rx = 0.8 mm, ry = 0.3
mm, rz = 0.9 mm in the reference frame of the vision system, where z is the depth
variable, y is the vertical axis.
It may be observed that repeatability of the x coordinate is worse than that of the y
coordinate, in spite of both being the image plane coordinates. This could be attributed
to the fact that the camera’s axis was only slightly inclined with respect to the table
surface, thus y coordinates of points of both line segments vary much less than x and z
coordinates.
Angular repeatability of the acquired reference frame is a separate issue to consider.
We calculated the average direction of the X axis and calculated angular misalignment
for every sample. It appears that on average the angle between the calculated and the
average direction of the X axis is 0.13° with the standard deviation of 0.08°. The
Fig. 10. Histograms and kernel density estimations of the angular misalignment of the
calculated reference frame with respect to average results.
histogram and the kernel density estimation are displayed in Fig. 10. This order of
magnitude of the angular repeatability implies that transverse errors at 1 m away from
the origin may be as high as 2.3 mm.
To classify workpieces, transfer learning of the AlexNet convolutional neural
network was used for the new data set. A new data set has been divided into five classes
with one more for the outliers. The training model is composed by 23 fully connected
layers for 6 class of workpieces and 25 layers for classification. The data are split 80%
for training and 20% for validation. After training the model, the verification set was
identified with 94.14% hit rate. This is made possible because all the training objects
use the same background, there aren’t disturbances and lights are diffuse so that there
aren’t shadows. Furthermore, on the table there is only one object per time. Figure 10
shows the training outcomes in terms of accuracy and loss function.
Accuracy is the ratio between the number of the correct results, and the total
number of samples:
Ntp þ Ntn
accuracy ¼ ð1Þ
Ntp þ Ntn þ Nfp þ Nfn
where N is the total number of samples, while N with indices tp, tn, fp, fn indicate the
number of true positive, true negative, false positive and false negative results
respectively.
Loss function is the cost rate for sample with weights w derived from AlexNet
(Fig. 11):
Nfp Nfn
loss ¼ wfp þ wfn ð2Þ
N N
Fig. 11. Training results: accuracy and loss indicators.
As some industrial workpieces happen to be so similar that can be distinguished

only by their code number, reliable workpiece identification in a factory application
should be better performed with the help of a coding system, like a barcode or a printed
code.
6 Conclusions
Robotic application of Industry 4.0 enabling technology needs easy and standardized
interfaces with built in diagnostic devices to make possible service and support. It is
apparent that robot operations increasingly depend on information generated by sen-
sors, worker instructions and/or CAD product data.
In the experiment, state of the art vision application and an efficient robot learning
procedure allows to reduce the need of accurate positioning of the workpiece under the
tolerances provided by today’s AMR without the need of jigs and fixtures and therefore
allowing an unprecedented flexibility of task executions.
The actual industrial application of the technology will depend on the simplicity of
its use and on its reliability and this aspect still needs non-trivial investigations.
References
1. Villani, V., Pini, F., Leali, F., Secchi, C., Fantuzzi, C.: Survey on human-robot interaction
for robot programming in industrial applications. IFAC-PapersOnLine 51(11), 66–71 (2018)
2. Rosenstrauch, M.J., Krüger, J.: Safe human-robot-collaboration-introduction and experiment
using ISO/TS 15066. In: IEEE Control, Automation and Robotics (ICCAR), pp. 740–744
(2017)
3. Bostelman, R.V., Hong, T.H.: Mobile manipulator stability measurements (No. Technical
Note (NIST TN)-1955) (2017)
4. Cheng, H., Chen, H., Liu, Y.: Topological indoor localization and navigation for
autonomous mobile robot. IEEE Trans. Autom. Sci. Eng. 12(2), 729–738 (2015)
5. Allmacher, C., Dudczig, M., Klimant, P., Putz, M.: Virtual prototyping technologies
enabling resource-efficient and human-centered product development. Proc. Manuf. 21, 749–
756 (2018)
6. Ma, C., Gao, H., Ding, L., Tao, J., Xia, K., Yu, H., Deng, Z.: Optimal energy consumption
for mobile manipulators executing door-opening task. Math. Probl. Eng. (2018)
7. Bayona, A., SanMiguel, J., Martinez, J.: Comparative evaluation of stationary foreground
object detection algorithms based on background subtraction techniques. In: Sixth IEEE
International Conference on Advanced Video and Signal Based Surveillance, AVSS 2009,
pp. 25–30 (2009)
8. Gould, S., Gao, T., Koller, D.: Region-based segmentation and object detection. In: Advances
in Neural Information Processing Systems, pp. 655–663. Curran Associates (2009)
9. Borji, A., Sihite, D.N., Itti, L.: Salient object detection: a bench mark. In: Computer Vision -
ECCV 2012, pp. 414–429. Springer (2012)
10. Bi, Z.M., Wang, L.: Advances in 3D data acquisition and processing for industrial
applications. Robot. Comput.-Integr. Manuf. 26, 403–413 (2010)
11. Makris, S., Michalos, G., Chryssolouris, G.: RFID driven robotic assembly for random mix
manufacturing. Robot. Comput.-Integr. Manuf. 28, 359–365 (2012)
12. Babic, B., Nesic, N., Miljkovic, Z.: A review of automated feature recognition with rule-
based pattern recognition. Comput. Ind. 59, 321–337 (2008)
13. Viola, P., Jones, M.J.: Robust real-time face detection. Int. J. Comput. Vis. 57, 137–154
(2004)
14. Astanin, S., Antonelli, D., Chiabert, P., Alletto, C.: Reflective workpiece detection and
localization for flexible robotic cells. Robot. Comput.-Integr. Manuf. 44, 190–198 (2017)
15. Stadler, S., Kain, K., Giuliani, M., Mirnig, N., Stollnberger, G., Tscheligi, M.: Augmented
reality for industrial robot programmers: workload analysis for task-based, augmented
reality-supported robot control. In: 2016 25th IEEE International Symposium on Robot and
Human Interactive Communication (RO-MAN), pp. 179–184. IEEE, August 2016
16. Song, D., Ek, C.H., Huebner, K., Kragic, D.: Task-based robot grasp planning using
probabilistic inference. IEEE Trans. Robot. 31(3), 546–561 (2015)
17. Mohseni-Kabir, A., Rich, C., Chernova, S., Sidner, C. L., Miller, D.: Interactive hierarchical
task learning from a single demonstration. In: Proceedings of the Tenth Annual ACM/IEEE
International Conference on Human-Robot Interaction, pp. 205–212. ACM, March 2015
18. Gottipolu, R.B., Ghosh, K.: A simplified and efficient representation for evaluation and
selection of assembly sequences. Comput. Ind. 50(3), 251–264 (2003)
19. Bruno, G., Antonelli, D.: Dynamic task classification and assignment for the management of
human-robot collaborative teams in workcells. Int. J. Adv. Manuf. Technol. 98(9–12), 2415–
2427 (2018)
Simulation Modeling of Assembly Processes
for Digital Manufacturing
Sławomir Kłos(&) and Justyna Patalas-Maliszewska
Faculty of Mechanical Engineering, University of Zielona Gora,

Licealna 9, 65-417 Zielona Gora, Poland
s.klos@iizp.uz.zgora.pl
Abstract. Computer simulation plays an important role in the modelling and

analysis of behavior-discrete manufacturing systems. The concept of Industry
4.0 includes computer simulation as a method for building Cyber Physical
Systems (CPS). The important problem that appears when building simulation
models is how, exactly, should the simulation model reflect real manufacturing
systems. In this paper, a case study of the computer simulation of assembly
processes is presented. The simulation model, presented in the paper, is prepared
using Tecnomatix Plant Simulation software, based on the design concept of a
real manufacturing system. The procedure for building and analysing a simu-
lation model, in order to obtain an efficient rate for a manufacturing system, is
proposed.
Keywords: Computer simulation Assembly processes

Efficiency of a manufacturing system Industry 4.0 Cyber-physical system
1 Introduction
Digital factories play important role in the Industry 4.0 (I4.0) concept, with computer
simulations being an integral part of that concept. ‘Digital Twins’, so-called and the
adaptation of manufacturing systems to a portfolio of production orders, by simulating
different configuration variants will be the answer, in the future, to the massive cus-
tomisation of products [1]. The CPS concept is a very important component of the next
generation of industry (I4.0), in which computer simulation will facilitate the modelling
of technological operations, will interact with real manufacturing systems and will
reflect them [2]. The significant role in the industrial engineering of the simulation
method is presented in many research works [3–6]. Computer simulation techniques
are often used for the study of manufacturing systems such as [7]:
• general system design and facility design/layout,
• material handling system design,
• cellular manufacturing system design,
• flexible manufacturing system design,
• manufacturing operations, planning and scheduling,
• maintenance operations, planning and scheduling,
• real-time control,

https://doi.org/10.1007/978-3-030-18715-6_22
262 S. Kłos and J. Patalas-Maliszewska
• operating policies,
• performance analysis.
Xiong et al. proposed development of the virtual reality technology, where users can
perform interactive assembly tasks by direct manipulation in virtual environment. The
user completes an assembly task by establishing the connection relationship between
objects, which can be implemented based on the object’s static interaction behavior [8].
Park proposed is an object-oriented methodology to create a virtual flexible manu-
facturing system (FMS) model. For the implementation of the proposed virtual FMS
model, the author uses Discrete Event Systems Specifications (DEVS) formalism,
which supports the specification of discrete event models in a hierarchical, modular
manner [9]. Implementation of simulation methods- for improving efficiency in a
manufacturing system- is often the objective of research work [10–12]. Computer
simulation is a highly effective tool for visualising, understanding and analysing the
dynamics of manufacturing systems. To simulate production processes in real time, a
simulation model of the manufacturing system should be prepared. It is, however, a
most important, expensive and time-consuming process. In Fig. 1, a procedure for
modelling the simulation of a discrete manufacturing system is proposed. The proce-
dure comprises seven steps. In the first step, a quantitative description of the manu-
facturing system should be given. The description should include details of the number
1. A quantitative description of the manufacturing system.
2. Determining the accuracy of the simulation model.
3. Design of the model using selected software.
4. Definition of simulation experiments and monitoring values.
5. Conducting the simulation experiments.
6. Validation the results and verification of the model.
Is the model No
accurate?
Yes
7. Determine the parameters for an actual manufacturing system.
Fig. 1. Procedure for modelling the simulation of a manufacturing system

Simulation Modeling of Assembly Processes 263
of manufacturing resources, buffers, the topology of the system, processing and setup
times, the production programme, etc. In the next step, accuracy of the description of
the manufacturing system should be established, giving details, for example, as to how
the material flow will be reflected viz. by simple connections between machines,
conveyors, forklifts, automated guide vehicles, etc. In the third step the design of the
simulation model should be prepared and the first test simulation could be carried out
with it. The input and output values for the simulation experiments should be proposed.
The number of experiments, the number of observations for each experiment and the
range of input values should be the next to be defined. The run time of the experiment
depends on the complexity of the system and the prescribed time, for example 8 h per
experiment. If the data for the simulation experiments is ready, the simulation process
can be run and the results of experiments should be able to be registered. The set of
input values can be changed dependent on the results. If the assumed results are
reached, the parameters can be implemented in a real manufacturing system, if not, the
model should be verified. Today, simulation software includes a plethora of very
sophisticated tools for analysing behavioural systems as neural networks, or as genetic
algorithms, for learning and for predicting the behavior of a system. In the paper, the
application of neural networks, for simulating analyses of computer results, will be
presented. In the next chapter the model of manufacturing system will be described.
2 Model of a Manufacturing System: A Case Study
Let us consider a manufacturing assembly system. The system includes 5 machines and
3 assembly stations. The simple structure of the system is shown in Fig. 2. Machines
M1 and M2 deliver parts for assembly station A1, while machines M3 and M4 deliver
parts for station A2. Station A3’s assembly parts come from stations A1 and A2. M5 is
a packing machine. In the manufacturing system, six parts is machined. The processing
times on the machines and assembly stations, as well as the set-up times, are based on a
lognormal distribution. A lognormal distribution is a continuous distribution in which a
random number has a natural logarithm, corresponding to a normal distribution. The
realisations are non-negative, real numbers. The density of the lognormal distribution
Lognor (r, l) is calculated as follows:
" #
1 lnðx l0 Þ2
f ðxÞ ¼ pffiffiffiffiffiffi exp ð1Þ
r0 x 2p 2r20
where r and l are respectively mean and standard deviations and are defined as
follows:
h r0 i
l ¼ exp l0 þ ð2Þ
2
r2 ¼ expð2l0 þ r20 Þ ðexpðr20 Þ 1Þ ð3Þ

The maximum of the density function is defined as:
expðl0 r20 Þ ð4Þ
Fig. 2. Simple simulation Model S1 for the assembly of a manufacturing system
In Table 1, the processing and setup times for various machine parts (E1–E6) and
the assembly stations are presented. The parts are assembled onto semi-finished
products E1 and E4, respectively, at assembly stations A1 and A2 and the ready
product E1 is assembled at station A3. The model does not include the manufacturing
Table 1. Processing and setup times for machines and assembly stations
Products/machine Operation times Setup times
r2 l r2 l
E1/M1 480 20 300 10
E2/M2 280 50 300 10
E3/M2 580 40 300 10
E4/M3 180 50 300 10
E5/M4 180 20 300 10
E6/M4 380 50 300 10
E7/M4 280 20 300 10
A1/E1 200 50 300 10
A2/E4 400 50 300 10
A3/E1 200 50 300 10
M5 200 50 0 0
personnel (workplaces). The production program is determined by batch sizes, q.v.

Table 2. As the input data for the simulation experiments are chosen: the availability of
the machines and assembly stations and the capacity of buffers B1, B2, …B6. The total
capacity of the buffers is determined as B = B1 + B2 + …+ B6.
Table 2. Production programme for individual parts

Part/ Batch Part/ Batch Part/ Batch
warehouse size warehouse size warehouse size
E2/W2 5 E5/W4 5 E7/W4 15
E3/W2 10 E6/W4 5 E5/W4 15
E2/W2 10 E7/W4 5 E6/W4 10
E3/W2 15 E5/W4 10 E7/W4 5
E2/W2 5 E6/W4 10 E5/W4 10
E3/W2 5 E7/W4 10 E6/W4 15
E2/W2 15 E5/W4 5 E7/W4 5
E3/W2 20 E6/W4 10
To analyse the flow of material within the manufacturing system, the Sankey
Diagram is analysed, this being a function of Tecnomatix Plant Simulation software,
for the flow of material in a system.
An analysis of the Sankey Diagram of the flow of material within a system is
presented in Fig. 3. The flow of material in the system is regular and similar in the all
branches of the system structure. The results of the analysis of the parameters given,
indicate that the resources of the system are loaded evenly.
Fig. 3. The Sankey Diagram analysis of Model S1

As with the input values for the simulation experiments, the availability of man-
ufacturing resources and buffer capacities are taken into account.
Simulation experiments are prepared for the values of the availability of resources
in the range from 95% to 99%. The range of buffer capacities is from 1 to 5.
In the first step, 50 random simulation experiments for the assumptions are gen-
erated. Examples of the first 10, randomly generated, input values, for simulation
experiments, are presented in Table 3.
Table 3. Examples of randomly generated, input values of the simulation experiments

Experiment M1 M2 M3 M4 M5 A1 A2 A3 B1 B2 B3 B4 B5 B6
Exp 06 97 95 99 98 98 98 97 96 4 1 1 2 1 3
Exp 07 97 97 98 96 96 96 99 99 5 1 4 5 1 4
Exp 08 96 96 97 96 99 95 97 96 5 1 1 5 2 1
Exp 09 99 95 99 95 95 95 97 98 1 2 4 2 1 5
Exp 10 99 96 98 96 95 98 98 98 3 1 2 1 1 4
Exp 11 96 96 96 95 99 96 98 98 5 1 4 2 5 3
Exp 12 98 95 96 97 95 99 97 97 3 1 3 2 5 1
Exp 13 96 97 97 99 98 97 98 99 5 1 2 5 1 4
Exp 14 97 96 95 99 99 96 99 97 1 2 5 1 2 2
Exp 15 98 97 98 98 98 97 97 95 4 1 1 4 2 4
The simulation experiments are conducted and arranged according to the system
throughput, from the smallest to the largest throughput values. Next, the input values of
the 10, manually defined experiments are added (see Table 4), where the availability of
resources was 95% or 99% and the capacity of all the buffers was 1, 2, 3, 4 or 5.
Manually defined input values, with an availability of 95%, are designated as first
experiments while those with an availability of 99%, are designated as last experiments.
Table 4. Examples of randomly generated, input values of simulation experiments

Exp 01 95 95 95 95 95 95 95 95 1 1 1 1 1 1
Exp 02 95 95 95 95 95 95 95 95 2 2 2 2 2 2
Exp 03 95 95 95 95 95 95 95 95 3 3 3 3 3 3
Exp 04 95 95 95 95 95 95 95 95 4 4 4 4 4 4
Exp 05 95 95 95 95 95 95 95 95 5 5 5 5 5 5
Exp 56 99 99 99 99 99 99 99 99 1 1 1 1 1 1
Exp 57 99 99 99 99 99 99 99 99 2 2 2 2 2 2
Exp 58 99 99 99 99 99 99 99 99 3 3 3 3 3 3
Exp 59 99 99 99 99 99 99 99 99 4 4 4 4 4 4
Exp 60 99 99 99 99 99 99 99 99 5 5 5 5 5 5
The number of manual and randomly defined input values, totals 60 experiments.
The results, the analyses and a discussion of the simulation experiments are presented
in the next chapter.
3 The Results of the Simulation Experiments
The simulation experiments are conducted using Tecnomatix Plan Simulation software
v. 12.2.0. Each experiment includes 3 observations and is completed over 5 simulation
days. The time taken to simulate 60 experiments is about 40 s. The throughput values
and the average lifespan of products (Li where i – number of experiment) in individual
experiments, are presented as a chart in Fig. 4. The first 5 experiments show us that
buffer capacity has a strong impact on the throughput of the system. The greatest
throughput value, -viz., Th60 = 7,31 products per hour- is reached in the last experi-
ment, with an availability of resources of 99% and a maximum, buffer capacity of 5.
The result of experiment 5, where throughput is Th05 = 6,99 products per hour, is very
interesting, since the availability of all resources is 95% (the lowest) and the buffer
capacity is at the maximum. This means that, by the proper allocation of buffer
capacity, the system’s throughput could be on the same level as Exp 29, where the
average availability of resources is 97,25%. The small buffers, equal to 1 (experiment
56), determine the relatively small value of the throughput (Th56 = 6,81 products per
hour) where the availability of all resources is 99%. A similar throughput value is
reached for experiment Exp 20 where the average availability of resources is 96,75%.
Fig. 4. Throughput values, per hour, for Model S1 simulation experiments.
In Fig. 5, the average lifespan of products, for individual simulation experiments, is

presented - ‘average lifespan’ meaning the average time that a product remains in the
system.
Fig. 5. Average lifespan values of products for Model S1 simulation experiments.
The smallest lifespan value (L = 0:39:04) is reached for experiment Exp 44 where
the average availability of resources is 96,88 and the total buffer capacity of the system
is 6. To find the relation between throughput and average lifespan, production flow
index h, is proposed where
h ¼Thi=Li ð5Þ
Flow index values for the simulation experiments, are presented in Fig. 6.
Fig. 6. Production flow index values for Model S1 of the manufacturing system.
In the next chapter, a more detailed model of a manufacturing system will be

analysed.
4 Changing the Complexity of the Model
Model S1 is very simple. Manufacturing workers, for example, are not taken into
account and the flow of material is reflected only generally. In Fig. 7, a new model, S2,
for the manufacturing system, is proposed. For each manufacturing resource, a
workplace is set aside, since manufacturing workers are required to complete opera-
tions on a machine. In the system, 8 workplaces (W1, W2,…, W8) are defined.
Additionally, the conveyors that deliver the parts to the machines and assembly stations
are allocated in the system. Changing the complexity of the model enables the impact
of the number of employees within the system’s throughput, as well as average
lifespan, to be analysed. The order of the simulation experiments is not changed;
experiments are conducted for 3, 5 and 7 manufacturing workers.
Fig. 7. Complex simulation Model S2, for the assembly of a manufacturing system
In Fig. 8, the throughput for different numbers of employees is presented. Gener-

ally, the system’s throughput increases pro rata, with any increase in the number of
manufacturing workers. Variability- that is, the impact of buffer capacity and the
availability of resources- is lower for smaller numbers of employees.
In Fig. 9, the average lifespan of a product in the model S2 of the manufacturing
system, is presented. The largest lifespan, that is, work in progress, is for the system
with 3 workers. For the model with 5 and 7 workers, the difference is negligible.
In the last step, the production flow index is analysed (Fig. 10). For 3 workers, the
value of the production flow index (h 50) is approximately 50.
Fig. 8. Throughput values, per hour, for Model S2 simulation experiments, for varying numbers
of manufacturing workers.
Fig. 9. Average lifespan values of a product, for Model S2 simulation experiments, for varying
numbers of manufacturing workers.
An increase in the number of production workers impacts on the variability of the

index and increases its value. Model S2, as proposed, could be a cyber twin of a real
manufacturing system. The efficiency of the system can be analysed before each
change to a production program. Based on the results of simulation experiments, a
database could be created for such a system. Such a knowledge database could be the
basis of a decision support system for establishing the parameters of the assumed
efficiency of the manufacturing system.
Fig. 10. Production flow index values for Model S2 of the manufacturing system, for varying
numbers of manufacturing workers.
5 Conclusions
In the paper the procedure for modelling the simulation of a manufacturing system is
proposed. The model is created firstly, as a simple simulation of the system. The set of
parameters and the simulation experiments are defined. After analysing the results, the
model is verified. In the next, more complex model which includes manufacturing
workers and the logistics infrastructure, the simulation experiments are repeated. The
simulation experiments are conducted for the different allocation of intermediate, buffer
capacities and the availability of manufacturing resources. The system throughput and
the average lifespan of the products are analysed as output values. The following
conclusions can be formulated:
• the single buffer capacities results with relatively small value of the system
throughput in the model S1 of the system,
• the allocation of buffer capacity and the availability of manufacturing resources has
a significant impact on the effectiveness of a manufacturing system,
• for different allocation of buffer capacity, the impact of the availability of manu-
facturing resources on the effectiveness of the manufacturing system can be
changed,
• the production flow index enables a compromise between the relatively high
throughput of a system and the low level of work in process to be achieved - q.v. the
short, average lifespan of products,
• it is not significant difference between the average lifespan for 5 ad 7 workers in the
model S2 of the manufacturing system,
• the highest values of the production flow index (greater than 200) is reached for the
experiments: Exp 44 (h 212), Exp 47 (h 207) and Exp 49 (h 207) (see
Table 5).
Table 5. The experiments generated the highest values of the production flow index
Exp 44 96 98 96 96 99 98 96 96 1 5 2 1 1 5
Exp 47 98 99 97 99 98 97 96 97 1 4 3 5 1 5
Exp 60 95 98 98 98 97 98 95 97 3 5 5 3 4 2
Experiments were conducted for different numbers of manufacturing resources. An

increase in the number of workers results in an increase in throughput and also in the
variability of the production flow index. The proposed simulation models can be used
as a part of a Cyber Physical System where the results of simulation experiments could
support operational decisions for production, planning and control. In further research,
the methods and tools of artificial intelligence, as neural networks and genetic algo-
rithms will be used to improve the total effectiveness of manufacturing systems [13]
and the production flow index. The presented study could be a base for building a cloud
simulation platform and the associated services that will provide the computing
resources and services for hybrid simulation of Virtual Manufacturing Systems [14].
References
1. Negri, E., Fumagalli, L., Macchi, M.: A review of the roles of digital twin in CPS-based
production systems’. Proc. Manuf. 11, 939–948 (2017). https://doi.org/10.1016/j.promfg.
2017.07.198
2. Varela, M.L.R., Putnik, G.D., Manupati, V.K., Rajyalakshmi, G., Trojanowska, J., Machdo,
J.: Collaborative manufacturing based on cloud, and on other i4.0 oriented principles and
technologies: a systematic literature review and reflections. Manage. Prod. Eng. Rev. 9(3),
90–99 (2018). https://doi.org/10.24425/119538
3. Polenghi, A., Fumagalli, L., Roda, I.: Role of simulation in industrial engineering: focus on
manufacturing systems. IFAC PapersOnLine 51–11, 496–501 (2018)
manufacturing systems’. Manuf. Lett. 3, 18–23 (2015). https://doi.org/10.1016/j.mfglet.
2014.12.001. Society of Manufacturing Engineers (SME)
5. Saldivar, A.A.F., Li, Y., Chen, W., Zhan, Z., Zhang, J., Chen, L.Y.: Industry 4.0 withcyber-
physical integration: a design and manufacture perspective. In: 2015 21st International
Conference on Automation and Computing (ICAC), pp. 1–6. IEEE (2015). https://doi.org/
10.1109/IConAC.2015.7313954
6. Chryssolouris, G., Mavrikios, D., Papakostas, N., Mourtzis, D., Michalos, G., Georgoulias,
K.: Digital manufacturing: history, perspectives, and outlook. Proc. Inst. Mech. Eng. Part B:
J. Eng. Manuf. 223, 451–462 (2009). https://doi.org/10.1243/09544054JEM1241
7. Negahban, A., Smith, J.S.: Simulation for manufacturing system design and operation:
literature review and analysis. J. Manuf. Syst. 33, 241–261 (2014)
8. Xiong, W., Wang, Q.H., Huang, Z.D.: A framework for interactive assembly task simulation
in virtual environment. Int. J. Adv. Manuf. Technol. 85(5), 955–969 (2016)
9. Park, S.C.: A methodology for creating a virtual model for a flexible manufacturing system.
Comput. Ind. 56(7), 734–746 (2005)
10. Trojanowska, J., Kolinski, A., Galusik, D., Varela, M.L.R., Machado, J.: A methodology of
(2018)
11. Straka, M., Trebuňa, P., Rosová, A., Malindžáková, M., Makysova, H.: Simulation of the
process for production of plastics films as a way to increase the competitiveness of the
company. Przemysl chemiczny 95(1), 37–41 (2016). ISSN 0033-2496
12. Kłos, S., Patalas-Maliszewska, J.: An analysis of the efficiency of a parallel-serial
manufacturing system using simulation. In: Intelligent Systems in Production Engineering
and Maintenance, ISPEM 2018. Springer (2019). Adv. Intell. Syst. Comput., 835, 32–43.
ISBN 9783319974897. https://doi.org/10.1007/978-3-319-97490-3_4
S., Jurczyk, M. (eds.) Advances in Manufacturing, pp. 131–140. Lecture Notes in
Mechanical Engineering. Springer, Cham (2018)
14. Dobrescu, R., Merezeanu, D., Mocanu, S.: Process simulation platform for virtual
manufacturing systems evaluation. Comput. Ind. 104, 131–140 (2019). https://doi.org/10.
1016/j.compind.2018.09.008
An Indicators Framework for Sustainability
Assessment of RFID Systems in Manufacturing
Bartłomiej Gładysz(&) and Aldona Kluczek
Faculty of Production Engineering, Institute of Production Systems Organization,

Warsaw University of Technology, Warsaw, Poland
bartlomiej.gladysz@pw.edu.pl
Abstract. Nowadays, manufacturing companies need to operate a business and

apply technology that are competitive as well as sustainable from a social,
environmental as well as economical perspective. One of the technologies
gaining substantial interests over manufacturing companies is radio frequency
identification, which supports logistics and information flow within organiza-
tions. Much of the literature and theory surrounding assessment of RFID sys-
tems explores environmental (mainly positive) and economic (cost and benefits)
impacts only. Some researchers have argued that current assessment methods
often fail to involve sufficient evaluation in terms of social considerations.
Therefore, an indicator-based sustainability assessment framework of RFID
technology should be expected. This paper aims to close this research gap in the
literature and present an indicators framework for sustainability assessment of
RFID systems in manufacturing. Results depict a list of designed RFID indi-
cators which, depending on the complexity of RFID technology to be imple-
mented may convenient for assessing sustainability of the technology. It
concludes based on the analyzed literature that the identification of RFID sys-
tems’ industrial indicators lies in accessibility and reliability data used for their
validation. Finally, the indicator-based framework for sustainability assessment
is verified for its usefulness on the case of dynamic spaghetti diagrams sup-
ported by Real Time Locating System. The paper concludes with a brief dis-
cussion on opportunities and limits of RFID-dedicated sustainability indicators.
Keywords: RFID Radio frequency identification Sustainability assessment

Technology assessment
1 Introduction
Sustainability concern has exponentially risen as an important topic within the

emerging technology and its assessment. RFID enjoys extensive investor confidence
and is poised for growth due to the relatively large range of its applications, e.g. in
transportation/distribution, health care systems [1]. Moreover, RFID is being intro-
duced in many industrial sectors [2] due to the logistic advantages related with the
utilization of these electronic devices. Numerous authors report successful RFID
deployments and benefits in many areas e.g. production logistics [3] warehousing [4],
external supply chains [5].

https://doi.org/10.1007/978-3-030-18715-6_23
An Indicators Framework for Sustainability Assessment 275
RFID is an automatic identification technology which is already used in the supply

chain management of many companies but is also gradually applied to the core of
manufacturing processes/adopting on the shop floors in production optimization, e.g.
less incorrect machining, decrease of lead times, [6–8], less inventory shrinkage. RFID
systems use radio waves to identify items automatically. These systems consist basi-
cally of RFID tags, transceivers (readers) and antennas. Passive RFID tags work on the
basis that they absorb the power from the reader and use this to empower the microchip
and re-emit a signal. Active tags emit signal themselves, therefore they always must be
connected to power supply (usually a battery), which powers the chip and transmits
identifier to the reader [9]. For detailed description of physics and applications authors
refer to the study [10].
Application areas of RFID systems are numerous and continually evolving, what
has impact on long-term corporate strategies, global markets, and its potential role for
sustained corporate growth and profitability [1, 8]. The advantage of RFID systems
compared to other auto-ID systems, such as the commonly used barcode, is that
multiple objects can be detected automatically and simultaneously without visual
contact. However, certain materials in the environment, such as metal, produce a
shielding effect.
More importantly, there is some debate over whether RFID tags are truly good for
the environment or and have an impact on the health. RFID technology falls under
Restriction of Hazardous Substances and Services (RoHS 2002/95/CE) regulation
regarding environmental issues. Waste created by tags presents problems during the
recycling process [11]. They are made of materials that are hard to recycle and may
present similar health risks as cell phones. The presence of these devices in e.g. plastic
packaging, and consequently in plastic waste, can cause several problems in the
recycling plants due to the materials included in these devices [12].
What indicators should be used to assess RFID technology in term of sustainability
considering also social impact?
How environmental, social and economic information is analyzed, integrated and
presented to decision-makers/practitioners is the detailed concern of sustainability
assessment?
Answering these questions does not give decision makers much information about
the real effects on climate change and social impact associated to the introduction of
RFID technology assessment in the industry/market, but should be the first approach to
the topic of RFID sustainability assessment. There is vivid discussion about privacy
and security concerns related to RFID [13]. There is no agreed methodology for
assessing RFID technology and trade off and synergistic impact on the sustainability.
RFID creates environmental and economic benefits, such as decreased shrinkage and
inventory, but it is also a source of waste and investment cost. Additionally, social
impact of RFID cannot be omitted, especially when considering tagging finished goods
delivered to consumers (e.g. apparel). Moreover, this is one of the most serious gaps in
the current technology assessment system—especially for large-scale technologies
where strategic risk and cost-sharing at the demonstration stage is unavoidable.
However, RFID metrics are limited due to factors such as: data availability and reli-
ability, no skills for measuring industrial sustainability progress with sustainability
assessment, there is the need to integrate environmental, economic and social issues
276 B. Gładysz and A. Kluczek
and seek their synergistic impact on the sustainability of RFID. For the purpose of this
paper, a contemporary approach to sustainability assessment framework using indi-
cators will be discussed. Sustainability assessment is being increasingly viewed as an
important tool to aid in the shift towards sustainability seeking to achieve the triple-
bottom-line objectives (environmental, economic, social) [14].
As a novelty is proposed a conceptual framework for assessing RFID system which
includes a three-tier structure of indicators, prior in-depth analyzing the existing lit-
erature on this technology for evaluating the applicability of RFID in the industrial
context. Due to an evident character of RFID systems and interactions with the eco-
nomic, physical environment, and social impact they are needed to investigate how
industry implements and assesses RFID systems. The framework will include the key
sustainability indicators comprising the fact that are both important for meeting sus-
tainability principles and relatively easy to collect. Hence, the goal is to depict a case
study for the usefulness of the indicators for industrial sustainability assessment of
RFID in terms of economic, environmental and social concerns.
The aim of this study is to propose possible indicators and associated metrics for
sustainability that can be used to assess the sustainability impact of RFID systems in
the field of manufacturing. Although RFID technology enables the tracking of
appropriate products and objects in real time since there are no clear and effective
measures to assess the impact in terms of environmental, economic, and social con-
cerns. For example, eco-efficiency is not clear due to its variants (i.e. environmental
intensity of production and environmental cost-effectiveness, environmental produc-
tivity, environmental improvement costs), application levels and improvement focuses,
if one attempts to evaluate both cost reductions and creating more value added to some
measure of environmental impact. This difficulty adds to the challenge for RFID
assessment which require exploratory research on industrial sustainability indicators
and associated metrics of RFID. Therefore, an indicators-based framework for sus-
tainability evaluation of RFID system in manufacturing will pose a challenge for
researchers and for practitioners.
2 State of the Art
Sustainability assessment methods based on environmental assessment dates back to

the pre-Brundtland era where sustainability mainly focused on environmental issues
such as resource consumption, pollution and impact. Currently available methodologies
for the assessment of the sustainability in production cover economic and environ-
mental parameters in a single integrated approach [15]. Other methods focus on one of
both topics only (economic or environmental considerations) [16, 17]. Life cycle
analysis (LCA) essentially identifies and quantifies the environmental effects caused by
a product over its whole life cycle, the potentiality of these effect and proposes envi-
ronmental impact reduction [18].
Two databases were browsed i.e. Web of Science and Scopus within keywords, for
the query:
<(“RFID” OR “Radio Frequency Identification”) AND “sustainability”>.
There were 53 papers found. The interest on the topic is growing (see Fig. 1).
However, after screening abstract no paper was selected, as no one treated about RFID
sustainability holistically. Either they relate to one specific sustainability aspect (e.g.
environmental [22], the impact (mostly positive) of RFID applications on sustainability
of the improved system (e.g. air luggage [19]) or they relate to the specific industry
(e.g. fresh food supply chain [20]) they relate to the sustainability of RFID system in
categories of its production, waste generated, energy consumed during production, etc.
(e.g. [21]).
Fig. 1. Number of papers on RFID sustainability in Web of Science and Scopus databases.
Study on relationship between radio frequency identification technology and

environmental sustainability found the application of RFID can enhance environmental
sustainability, by providing more accurate and timely information to an organization’s
green supply chain management practices [11, 22]. Kanth et al. [21] explored the
sustainability, environmental impacts and toxic effects in manufacturing process of the
RFID tag antenna, while the technology-organization-environment framework explains
the RFID system deployments for environmental sustainability [21, 23].
Other assessment methods particularly used in healthcare settings were developed
based on literature review for testing the impact of environmental factors on the quality
of RFID generated data [24]. Fisher and Monahan [25] concluded that RFID deploy-
ment in hospitals are closely affected by the staff concern of surveillance and that
medical staff labor is intensified because the task of keeping the systems operational
often falls upon them. Those social dimensions should be considered when designing
RFID system [25].
Unfortunately, standards providing requirements specifications on the implemen-

tation of RFID and on the sustainability assessment of RFID systems in manufacturing
are still missing. This underlines the need for detailed framework to assess the sus-
tainability of RFID system in which it will be deployed TBL objectives to be achieved.
With regard to the RFID system, none of the analyzed papers discussed holistic
approach taking into account:
• environmental benefits and contras,
• economic benefits and costs,
• social positive and negative impact,
Social dimension is still off and approached mainly from consumers’ privacy point of
view [13]. Hence, the theoretical framework will be proposed. It will aim to incorporate
insights into following areas from the triple-bottom-line perspective:
• what is an impact of RFID system on the performance of company’s processes?
• what is a sustainability of the system itself?
3 Framework
3.1 General Assumptions

This framework employs a wide range of indicators to focus on environmental, eco-
nomic and social impacts. Under it, the assessment of sustainability is considered as an
assessment of indicators by which RFID technology progress towards sustainability
can be tracked. The selection of the indicators considered here is driven by the spec-
ification of RFID technology discussed in Introduction as well as by the findings of
previous research on the sustainability of technology. The indicators are gathered in
three dimensions of TBL in terms of criteria domain. A relationship between criteria
domain, criteria and indicators, and sustainability dimensions is highlighted in Table 1.
In this study, the authors analyze sustainability indicators faced in manufacturing.
3.2 Indicators
A range of sustainability indicators that must be considered when assessing sustain-
ability of RFID systems is presented in a hierarchical structure (Table 1). It includes
non-traditional indicators, such as used in LCA analysis, but also indicators related to a
development way of RFID system/impacted by a development stage or a way of
production.
Table 1. The framework of indicators considering RFID system designed for manufacturing.
Criteria Indicators Metrics Unit
Economic Benefits Labour Decrease of cycle times (average) min
efficiency Lead time (average) decrease min
Increase of workers productivity vol./manhour
Decrease of errors no./manhour
Inventory Decrease of costs of frozen stocks USD/year
accuracy Decrease of stocks in monetary values USD
Decrease of costs of inventory handling USD/year
Decrease of shrinkage in monetary values USD/year
Decrease of costs of shrinkage handling USD/year
Costs Investment costs USD
MRO costs USD/year
Effectiveness ROI %
Payback period weeks
Environmental Waste Amount of Number of tags disposed completely pieces/year
generation waste Number of tags circulating in the system pieces
Electronic devices disposed completely pieces/year
Electronic devices installed pieces
Lifecycle & Tag’s lifecycle duration in supply chain weeks
supply chain No. of tag’s reads in its lifecycle pieces
integration No. of reading points in a supply chain no.
No. of supply chain echelons benefiting no.
RFID system
Waste Inventory Decrease of stocks in units pieces/g/m3
reduction accuracy Decrease of shrinkage in units pieces/g/m3
BIS
Assets Decrease of paper documents pages/year
utilization Decrease of printing accesories USD/year
Decrease of number of assets pieces
Decreae of the total value of assets USD/year
Decrease of fuel consumption in litres l/year
Decrease of electricity for transportation kWh/year
consumption
Social Public acceptability Qualitative assessment of customer’s scale 1-5
choice by users
TEM in place scale 1-5
Qualitative assessment of threats by users scale 1-5
Additional staff manhour/year
Relations improvement scale 1-5
Safety & health No. of occupational accidents no./manhour
Air pollutant kg CO2
3.2.1 Economic Impact Indicators

Economic benefits of RFID implementation were discussed widely in the literature
(e.g. [5, 26–29] and can be generalized to:
• improvement of supply chain transparency through actual and real time data what
enable better supply chain and inventory management (decrease of stocks),
• acceleration of product flow through automated, faster and error-proof identification
of goods as work in progress semi-finished and finished goods,
• decrease of labor costs related to logistics processes.
Several authors proposed framework for assessment of benefits that RFID can
provide from economical point of view e.g. simulation modelling [29, 30], analytic
models [27], reference models and process analysis [26, 31], return on investment
calculators [32], group decision making [33]. Those frameworks can be successfully
used for detailed analysis. From the point of view of sustainability assessment
framework, it is not important to focus on details of economic analysis, but to state
most important factors that could be estimated, evaluated and analyzed. Generally, it is
important to consider three main criteria domains in the economic sustainability
dimension i.e. improvements in terms of data accuracy, inventory levels and labor
effectiveness [3–8, 27, 29, 31]. Better data accuracy leads to the decrease of lead times
and shrinkage, which could be a quantitative indicator for this criterion. Additionally,
decrease of lead times and labor efficiency are indicators relating to the improvements
in terms of labor efficiency (identification operations eliminated or accelerated).
In terms of costs it is necessary to think about total cost of ownership, so invest-
ment, maintenance and repairs of hardware, software and services (especially instal-
lation) have to be considered.
When discussing economic dimension of sustainability, it is also necessary to use
cumulative indicators illustrating economic effectiveness. It is proposed to use basic
indicators as ROI and payback period, as they are mostly used by practitioners and
decision makers in companies.
3.2.2 Environmental Impact Indicators

Closed–loop systems, in which tags are circulating within chain, generate less waste. In
case of RFID, it is important because tags and readers are electronic devices, which are
made of different materials (plastic, metal, batteries). That makes them impossible to
recycle [12], very hard to dispose and almost impossible to remanufacture. Theoreti-
cally, in some cases readers and antennas could be, but it is rather not economically
justifiable. Closed-loop systems generate less need to buy new tags yearly. It is also
always better if tags are used in several echelons of supply chain than only locally,
because benefits could be achieved along the chain while the waste occurs only once. It
is proposed to use indicators related to the number of tags (circulating within the
system and being bought yearly) and number of devices in the system, as those indi-
cators depict the scope of the waste generated.
Authors propose to assess environmental impact of RFID systems from two con-
trary perspectives. First, RFID system is the source of new waste (e.g. tags are elec-
tronic waste). From the waste generation point of view, it should be assessed what
amount of waste is generated yearly (tags, readers, etc.) and what is the scope of the
system (number of circulating tags). It should be noted that RFID system components
may be exploited with different intensity. Therefore, it is proposed to assess also tags’
lifecycle duration and number of reads in a lifecycle. The highest values the more
intensive is exploitation, so the generated waste can deliver more economic benefits.
Second, RFID system implementation results in decrease of other wastes like shrink-
age, excessive inventory and transportation, low utilization of assets. Therefore, those
issues should be assessed. Such approach considering both positive and negative
environmental impact of RFID systems is missing in the literature and practice.
3.2.3 Social Impact Indicators

Social impact, both positive or negative is “(…) important to identify and quantify the
human risks and consequences will allow better acceptance and understanding of some
technologies that are often subject to public objection” [34]. The social dimension
includes air pollutions from energy consumption that can have adverse impact on
human health. It can measure the CO2 emissions from energy consumption of manu-
facturing of RFID systems. In this research, the social indicators were modified and
adopted based on the research of Santoyo-Castelazo and Azapagic to check their
applicability in RFID systems [35]. Other research applicable in a medical area
involved the social dimensions of RFID systems [25]. This social impact is in this work
measured by three indicators, covering the whole life cycle of RFID systems. The
social indicators that could be analyzed within the RFID sustainability assessment
decision-support framework are divided into three criteria categories: (i) public
acceptability; (ii) safety; and (iii) human health effects from external air quality.
Public acceptability can be expressed by providing information to consumers, so
they can make sustainable choices in their purchases and implementing technology
management system (TEM) in place. Safety aspect is mostly related to occupational
accidents (e.g. injuries affecting workers, illness rate) and accident performance along
the technology life cycle (e.g. explosions, oil spills, etc.) [35]. But in case of RFID
system, it should be also assessed in relation with privacy and surveillance threats,
which are specific problems when implementing RFID. Labor intensification increase
is another issue to be analyzed.
Air pollutant aspect has been quantified using the human toxicity potential esti-
mated within LCA. Emissions of air pollutant (AP) within the third criteria category,
are dispersed in the outdoor environment were estimated as in Eq. (1):
AP ¼ Ri ðEFi x Ei Þ ð1Þ
where:
• EF is the CO2 emission factor from energy i used for producing RFID expressed in
e.g. kgCO2/MJ;
• Ei is an energy contribution of process in MJ.
Estimation of AP is very difficult, as it would require participation of the systems’
manufacturer, systems’ integrator and systems’ user. Therefore, this indicator may be
rarely possible for application.
4 Results
The framework was verified for its usefulness on the case of dynamic spaghetti dia-
grams supported by Real Time Locating System [36]. Analytic Hierarchy Process [37]
was used as decision making method to estimate weights of criteria (see Table 1) and
ranking of alternatives by pairwise comparisons. AHP-OS application was used for
computational procedure [38]. Three alternatives [39] (Alt1- Passive UHF RTLS, Alt2
- Active Wi-Fi RTLS, Alt3 - Active UWB RTLS) were assessed by two experts
experienced in RFID deployments (10 years of managing, designing, analysis of over
50 deployments). For results of global priorities of criteria obtained using standard
linear scale, see Fig. 2. Consensus was very high (95.4%) and results were consistent
(consistency ratio: CRMAX = 8%). Those priorities were used to assess alternatives.
Fig. 2. Consolidated results of global priorities of criteria for RFID sustainability.
Results for comparisons of alternatives were compared for different scales and Alt3
was always ranked as the superior one, Alt2 as the second and Alt1 as the worst (see
Table 2). Only two of ten (power and adaptive) scales showed inconsistent result
(CRMAX > 10%). CRMAX for standard linear scale was 8.9%, so results may be treated
as consistent. Alt3 was not significantly preferable, but still achieved the highest score,
only for an inverse linear scale. The obtained findings confirm technology selection
results reported in [39].
Table 2. Ranking of alternatives prior to RFID sustainability.

Scale Group results CRMAX
Alt1 Alt2 Alt3
Standard linear [40] 23.2% 32.9% 43.9% 8.9%
Logarithmic [41] 26.1% 33.7% 40.2% 4.0%
Root square [42] 27.3% 33.9% 38.7% 2.2%
Inverse linear [43] 30.2% 34.0% 35.8% 0.2%
Balanced [44] 28.6% 34.1% 37.2% 0.5%
Balanced generalized [45] 27.6% 34.2% 38.2% 0.8%
Adaptive balanced [45] 26.8% 34.1% 39.0% 1.1%
Adaptive [45] 21.4% 31.7% 46.8% 15.6%
Power [42] 19.2% 28.6% 52.2% 36.6%
Geometric linear [46] 21.7% 32.5% 45.8% 5.6%
5 Conclusions and Discussion
The paper reviews the indicators-based sustainability assessment addressing the

research gap in the literature for expanding database of RFID sustainability indicators.
Results from the explanatory study on this field is to provide a list of RFID-
dedicated indicators which, depending on the complexity of the technology to be
implemented may be convenient for assessing sustainability. The list of indicators
could be used as representative of the whole picture of RFID-based projects. Desig-
nated indicators as a result of the field study could state a basis for sustainability
assessment of RFID-based projects. These indicators with metrics may also make
simple comparisons through analyzing the values of particular indicators. Due to a
widespread application of RFID, some benefits in manufacturing like real-time infor-
mation visibility and traceability are expected. The sustainability indicators considering
various factors that contribute to the success or failure of RFID systems in manufac-
turing should be further analyzed and applicability of these indicators should be val-
idated based on an industrial case study.
Although sustainability assessment indicators for RFID have been described by
various researchers and proposed by the authors (for implementing in various RFID-
based projects), effective selection methods for sustainability indicators (environmen-
tal, economic and social) are unavailable. A selection of appropriate indicators depends
on the size, the complexity, the area of projects in which RFID will be implemented.
Unfortunately, a great limitation belongs to the social indicators which rely on that the
most of social metrics are not quantitative, but qualitative in nature. The weights
assigned to them reflect the weights of the perceptions of expert groups.
The RFID – related sustainability indicators reported in this paper might be useful
for manufacturing facilities or practitioners Hence, sustainability assessment of RFID
requires considerable research efforts. The future research might be carried out by
enabling the combination of different methodologies into one integrated approach (e.g.
multi-criteria decision making) to select the number of relevant, more collaborative
indicators to be used for sustainability assessment of RFID in manufacture, as
presented in the Sect. 3. By participating the expert`s groups future research will
provide reliability of the analysis of a set of indicators, weights appointment, etc. From
other hand, an application of the selection methods allows ranking indicators, thus
making a high impact on the results.
It can conclude, based on the literature review that the problem in sustainability
assessment or RFID lies in gathering of reliable and accessible data (if indicators are
not qualitatively and insufficient) used for their validation focusing on real industrial
case studies.
Therefore, a comprehensive analysis of sustainability impact indicators selection
for RFID-based projects that consider the environment, economic and social values
should be done. It might provide meaningful information for decision makers should be
further discussed. In particular, it is a demanding as it involves various functional units
and the sustainability dimensions, and focus on diverse data collection methods and
generates a vast amount of data. As the manufacturing firms will have to manage real-
time tracking indicators.
References
1. Jung, K., Lee, S.: A systematic review of RFID applications and diffusion: key areas and
public policy issues. J. Open Innov.: Technol. Market Complex. 1(9), 1–19 (2015)
2. Turcu, C. (ed.): Sustainable Radio Frequency Identification Solutions. InTech (2010)
3. Zelbst, P.J., Green, K.W., Sower, V.E., Reyes, P.M.: Impact of RFID on manufacturing
effectiveness and efficiency. Int. J. Oper. Prod. Manag. 32(3), 329–350 (2012)
4. Osyk, B.A., Vijayaraman, B.S., Srinivasan, M., Dey, A.: RFID adoption and implementation
in warehousing. Manag. Res. Rev. 35(10), 904–926 (2012)
5. Sarac, A., Absi, N., Dauzére-Pérés, S.: A literature review on the impact of RFID
technologies on supply chain management. Int. J. Prod. Econ. 128(10), 77–95 (2010)
6. Fan, T.-J., Chang, X.-Y., Gu, C.-H., Yi, J.-J., Deng, S.: Benefits of RFID technology for
reducing inventory shrinkage. Int. J. Prod. Econ. 147, 659–665 (2014)
7. Lim, M.K., Bahr, W., Leung, S.C.H.: RFID in the warehouse: a literature analysis (1995–
2010) of its applications, benefits, challenges and future trends. Int. J. Prod. Econ. 145(1),
409–430 (2013)
8. Narsing, A.: RFID and supply chain management: an assessment of its economic, technical,
and productive viability in global operations. J. Appl. Bus. Res. 21(2), 75–80 (2005)
9. Gladysz, B.: Typology of RFID systems (2016). http://www.ptzp.org.pl/files/konferencje/
kzz/artyk_pdf_2016/T1/t1_0948.pdf. Accessed 18 Apr 2018
10. Finkenzeller, K.: RFID Handbook: Fundamentals and Applications in Contactless Smart
Cards and Identification. Wiley, New York (2010)
11. Sower, V., Bellah, J., Zelbst, P., Green, K.: RFID and Environmental Sustainability: An
Unlikely Partnership. RFID J. (2013). http://www.RFIDJorunal.com/articles/view?11008.
Accessed 18 Apr 2018
12. Aliaga, C., Ferreira, B., Hortal, M., Pancorbo, M.A., Lopez, J.M., Navas, F.J.: Influence of
RFID tags on recyclability of plastic packaging. Waste Manag. 31, 1133–1138 (2011)
13. Turri, A.M., Smith, S.W.: Kopp, privacy and RFID technology: a review of regulatory
efforts. J. Consum. Aff. 51(2), 329–354 (2017)
14. Elkington, J.: Accounting for the triple bottom line. Measuring Bus. Excell. 2(3), 18–22
(1998)
15. Wehbe, M.B., Seiler, R.A., Vianco, M., Mendoza, M.A., Baronio, A.M., Tonolli, A.J.:
Assessing sustainability of local production systems: a proposal based on socio-ecological
resilience and collaboration, GSDR 2015 Brief (2015). https://tinyurl.com/y8cvlava.
Accessed 18 Apr 2018
16. Hauschild, M., Jeswiet, J., Alting, L.: From life cycle assessment to sustainable production:
status and perspectives. CIRP Ann. – Manuf. Technol. 54(2), 1–21 (2008)
17. Hacking, T., Guthrie, P.: A framework for clarifying the meaning of triple bottom-line,
integrated, and sustainability assessment. Environ. Impact Assess. Rev. 28(2–3), 73–89
(2008)
18. Glavic, P., Lukman, R.: Review of sustainability terms and their definitions. J. Clean. Prod.
1, 1875–1885 (2007)
19. Wong, E.Y.C., Wong, W.H.: The development of reusable luggage tag with the internet of
things for mobile tracking and environmental sustainability. Sustainability 9(1), 58 (2017)
20. Bottani, E., Manfredi, M., Vignali, G., Volpi, A.: Life cycle assessment of RFID
implementation in the fresh food supply chain. Int. J. RF Technol.: Res. Appl. 6(1), 51–71
(2014)
21. Kanth, R.K., Liljeberg, P., Tenhunen, K., Wan, Q., Amin, Y., Shao, B., Chen, Q., Zheng, L.,
Kumar, H.: Evaluating sustainability, environmental assessment and toxic emissions during
manufacturing process of RFID based systems. In: IEEE 9th International Conference on
Dependable, Autonomic & Secure Computing, pp. 1066–1071 (2011)
22. Green Jr., K.W., Zelbst, P.J., Sower, V.E., Bellah, J.C.: Impact of radio frequency
identification technology on environmental sustainability. J. Comput. Inf. Syst. 57(3), 269–
277 (2017)
23. Angeles, R.: Using the technology-organization-environment framework and Zuboff’s
concepts for understanding environmental sustainability and RFID: two case studies. Int.
J. Econ. Manag. Eng. 7(11), 2878–2887 (2013)
24. van der Togt, R., Bakker, P.J.M., Jaspers, M.W.M.: A framework for performance and data
quality assessment of Radio Frequency IDentification (RFID) systems in health care settings.
J. Biomed. Inf. 44(2), 372–383 (2011)
25. Fisher, J.A., Monahan, T.: Tracking the social dimensions of RFID systems in hospitals. Int.
J. Med. Inf. 77(3), 176–183 (2008)
26. Attaran, M.: Critical success factors and challenges of implementing RFID in supply chain
management. J. Supply Chain Oper. Manag. 10(1), 144–164 (2012)
27. Dai, H., Tseng, M.M.: The impacts of RFID implementation on reducing inventory
inaccuracy in a multi-stage supply chain. Int. J. Prod. Econ. 139(2), 634–641 (2012)
28. Hardgrave, B.C., Aloysius, J., Goyal, S.: Does RFID reduce out of stocks? A preliminary
analysis. Int. J. RF Technol.: Res. Appl. 1(1), 44–56 (2009)
29. Ustundag, A., Tanyas, M.: The impacts of Radio Frequency Identification (RFID)
technology on supply chain costs. Transp. Res. Part E 45, 29–38 (2009)
30. Lee, I., Lee, B.-C.: An investment evaluation of supply chain RFID technologies: a
normative modeling approach. Int. J. Prod. Econ. 125(2), 313–323 (2010)
31. Becker, J., Weiss, L., Vilkov, B., Winkelmann, A.: A model based approach for calculating
the process driven business value of RFID investments. Int. J. Prod. Econ. 127(2), 358–371
(2010)
32. RFID Journal: ROI calculator (2009). https://tinyurl.com/ycrwqxsu. Accessed 18 Apr 2018
33. Chuu, S.J.: An investment evaluation of supply chain RFID technologies: a group decision-
making model with multiple information sources. Knowl.-Based Syst. 66, 210–220 (2014)
34. Evans, A., Strezov, V., Evans, T.J.: Assessment of sustainability indicators for renewable
energy technologies. Renewable Sustain. Energy Rev. 13, 1082–1088 (2009)
35. Santoyo-Castelazo, E., Azapagic, A.: Sustainability assessment of energy systems:

integrating environmental, economic and social aspects. J. Clean. Prod. 80, 119–138 (2014)
36. Gladysz, B., Santarek, K., Lysiak, C.: Dynamic spaghetti diagrams. A case study of pilot
RTLS implementation. In: Intelligent Systems in Production Engineering and Maintenance –
ISPEM 2017, pp. 238–248. Springer, Cham (2018)
37. Saaty, T.L.: Fundamentals of Decision Making and Priority Theory with the Analytic
Hierarchy Process. RWS, Pittsburgh (1994)
38. Goepel, K.D.: Implementation of an online software tool for the analytic hierarchy process
(AHP-OS). Int. J. Anal. Hierarchy Process 10(3), 469–487 (2018)
39. Gladysz, B., Santarek, K.: An approach to RTLS selection. DEStech Transactions on
Engineering and Technology Research (24th ICPR 2017), pp. 13–18 (2017)
40. Saaty, T.L.: The Analytic Hierarchy Process: Planning, Priority Setting, Resource
Allocation. McGraw-Hill, New York (1980)
41. Ishizaka, A., Balkenborg, D., Kaplan, T.: Influence of aggregation and measurement scale on
ranking a compromise alternative in AHP. J. Oper. Res. Soc. 62, 700–710 (2010)
42. Harker, P., Vargas, L.: The theory of ratio scale estimation: Saaty’s analytic hierarchy
process. Manag. Sci. 33, 1383–1403 (1987)
43. Ma, D., Zheng, X.: 9/9 - 9/1 scale method of AHP. In: 2nd International Symposium on
AHP, vol. 1, pp. 197–202 (1991)
44. Salo, A., Hamalainen, R.: On the measurement of preferences in the analytic hierarchy
process. J. Multi-Criteria Decis. Anal. 6, 309–319 (1997)
45. Goepel, K.D.: Comparison of judgment scales of the analytical hierarchy process — a new
approach. Int. J. Inf. Technol. Decis. Making (2018). https://doi.org/10.1142/s021962
2019500044
46. Lootsma, F.: Conflict resolution via pairwise comparison of concessions. Eur. J. Oper. Res.
40, 109–116 (1989)
Automated Training of Convolutional
Networks by Virtual 3D Models for Parts
Recognition in Assembly Process
Kamil Židek1(&), Peter Lazorík1, Ján Piteľ1, Ivan Pavlenko2,

and Alexander Hošovský1
1
Department of Industrial Engineering and Informatics,
Faculty of Manufacturing Technologies with a Seat in Presov,
Technical University of Kosice, 1 Bayerova Street, 08001 Prešov, Slovakia
kamil.zidek@tuke.sk
2
Department of General Mechanics and Machine Dynamics,
Faculty of Technical Systems and Energy Efficient Technologies,
Sumy State University, 2 Rymskogo-Korsakova Street, 40007 Sumy, Ukraine
Abstract. One of the most monotonous activities in using convolutional neural

networks for image recognition is preparation of the learning data. It involves
creating samples (2D images of object) at different angles of view, different
backgrounds/materials and partial overlay of the object. Input data must include
a relatively large number of frames, typically about 100 and more images per
object to make the learning precision useful. In the paper there is proposed a
new approach to creating these data fully automated based on a virtual 3D model
of the standardized parts. Automation principle is generating 2D images from
the imported 3D construction model, including the following variable parame-
ters: the angle of rotation, background and the material of the component. We
used for verification pretrained DNN model Faster RCNN Inception v2 with
single shot detection (SSD). The learned convolutional network was next tested
by real samples to verify a new approach of learning by virtual models and
recognition of real objects (parts).
Keywords: Deep learning Neural networks Image recognition 3D models
1 Introduction
The new trends in analyzing of 2D images (mainly object detection) are based on deep
learning techniques with combination of convolutional neural networks models. The
main difference between standard artificial neural networks (ANN) and convolutional
neural networks (CNN, or ConvNet) is different structures of layers with hidden
neurons and large numbers of neurons inside of layers. The standard ANN can be used
more generally but data preparation to input layers is much more complicated.
Convolutional neural networks can simplify the basic image processing approach,
mainly complicated process of extracting data from image data that has been done
previously through standard image recognition algorithms (thresholding, contouring,
segmentation, etc.) [1, 2]. By using selected convolutional model (RCNN, MobileNet,

https://doi.org/10.1007/978-3-030-18715-6_24
288 K. Židek et al.
Alexnet, etc.), it is possible to directly insert the raster images of objects to the neural
network inputs as training samples, which we want to recognize by neural network.
Convolutional networks have the highest recognition precision if we provide for
training process large quantity of samples, including multiple detections in a single
image and partially overlay image parts with other object (assembly or construction
nodes) [3, 4]. This way the learned CNN model can recognize partially visible objects
(partially mounted in assembly) and recognize from any view. These features give
CNN preconditions for the usability in recognizing the industrial machine parts
themselves or the individual parts at the construction nodes with support for Industry
4.0 concept in virtual or augmented reality devices. Nowadays manual assisted
assembling by augmented reality is current trend for production, where is not possible
to establish standard automation with production lines and industrial robots. The
assisted assembly process needs smart glasses to teach and check new workers in very
short time. The key feature for this process is reliable machine parts localization and
recognition in real-time. These requirements can be realized with deep learning of
convolutional networks.
If we have very variable production (many variants of products, small production
batch) the program preparation takes a long time (mainly preparation of samples for
parts training in CNN) and assisted assembly process is nonprofitable. One idea how to
make this process usable is teaching CNN model by virtual 3D models. Initial research
in the area of standardized machine parts recognition (hexagon screw/nut and circular
hole part element) by convolutional networks, with standard input data image collec-
tion with reliable results is described in [5].
The research works on testing the usability of a virtual 3D model as an input for
learning a convolutional network can be summarized as follow: research on training
and testing of learned detectors using virtual images [6], recursive analysis of 3D
models via convolution networks [7], training of CNN networks using 3D models [8],
learning of depth detectors from 3D models [9], 3D analysis of 3D objects via CNN
[10], CNN trained with 3D rendered images [11].
If we need connection with industrial 3D model it is necessary create bridge
between 3D design construction software and teaching software (CNN framework). It
is suitable select export format which is unique for all major 3D design software. Major
3D design software are: Pro Engineer Creo, Data Assault CATIA, Autodesk Inventor,
Solid Edge, Siemens NX, etc. The standard export formats for these 3D software are:
DXF, STL, VRML, OBJ, IGES, STEP, 3DS. The most precise format which is used for
rapid prototyping is STL format but removes any information about material or colour
from parts or assembly product. The most suitable format which holds information
about colours is WaveFront OBJ format. This format was primarily developed for 3D
animation software, but there is support in construction software too. There are many
research works concerning 3D models design, for example paper about creating 3D
design of assembly with many examples of case study [12], 3D cad lightweight rep-
resentation [13], enhancing parameters in 3D model [14].
Convolutional network works great for image object recognition, but there is still a
problem with manual preparation data for learning process. We need to collect or create
a large number of input 2D samples, usually several hundreds, which is a prerequisite
for its reliable learning of the object. If we need a large number of objects which we
Automated Training of Convolutional Networks by Virtual 3D Models 289
want to recognize, this introductory process can be extended to several weeks of

monotonous work. The object must be scanned in different angular and translation
variations with partial overlapping by other objects as well as with different back-
grounds and materials.
The use of 3D virtual models to train convolutional networks accelerates part of the
monotonous work, especially when creating variations and backgrounds, but it is still a
relatively tedious process. The article presents a new methodology for automated
generation of input data for learning without any monotonous manual work. Tasks such
as an angle/position definition, background and material change, object detection
position (SSD) can be automated by scripting language. In this way it is possible to
short the time of preparing samples (sufficient numbers of samples 100–300 per object)
for a few minutes by 3D model created previously in the construction software. The
methodology itself as well as the software implementation with the recognition veri-
fication is described in the following chapters.
2 Research Problem
2.1 Principle of New Approach to Teach CNN by Virtual 3D Model

The principle scheme of automated input data generation from a virtual 3D model for a
convolutional network is shown in the Fig. 1.
Detailed training data generation consists of several steps, which are fully
automated:
• Import of 3D model from any 3D CAD software (CATIA, CREO, Autodesk
Inventor, etc.) to open source 3D rendering software with scripting language.
• Generation of 2D images with part rotations/position, different textures and back-
grounds, including standard views.
• Generation of object bounding box in 2D image with parameters of position and
size for part localization by standard image processing technics (python OpenCV)
used for single shot detection (SSD).
• Generation of text description file (XML) with all image parameters: name, size and
part position.
• Creation a snapshot set with random parameters for learning validation: training
(80%), validation set (20%).
Import of 3D Model Genera on of 2D part images Iden fica on of part posi on with
from Design So ware to with different: rota on, standard image processing and data
script/render so ware posi on, material, background digitaliza on to XML file
Fig. 1. Scheme of sample generation from a 3D model.
The principle of rotation of objects in the scene is calculated through the object
rotation matrix with center point of gravity according the following formulas (1) and (2).

cos b 0 sin b 0 cos c sin c 0 0

0 1 0 0 sin c cos c 0 0
Ry ¼ ; R ¼ ð1Þ
sin b 0 cos b 0 z 0 0 1 0
0 0 0 1 0 0 0 1
where:
Ry, Rz - matrix of virtual 3D rotation of part,
b, c - angle rotation for each additional generated view.

1 0 0 0

0 1 0 0
Txyz ¼ ð2Þ
0 0 1 0
tx ty tz 1
where Txyz – matrix of virtual 3D translation of part; tx, ty, tz - angle rotation for each
additional slide.
We rotate object in two axes Y or Z at 20°, validation samples are generated with
random angle for rotation and random change of position.
2.2 Implementation Idea to Software Solution

The introduced solution presents automated script which is versatile for any 3D design
software that uses export to OBJ file format. We used for testing export from Autodesk
Inventor 2019.
We select Blender 3D software for rendering output images. This software is open
source and provides the Python internal language. We can directly control the current
scene by available libraries with this simple language.
We need to localize part inside of image, this task is realized by Python version of
OpenCV library. OpenCV detects width, height and X, Y position of part in generated
frame.
These data are next stored in XML file format which is generated for every frame
file (name of image file, type of object, image height/width in pixels, object position
and size in pixels).
TensorFlow framework (GPU variant) was selected for teaching pretrained model.
For experiments we selects Faster R-CNN Inception V2 model pretrained on coco
dataset. New data are trained by transfer learning. Learned frozen model is exported to
OpenCV for fast execution in embedded devices.
Detailed procedure for creating input data from virtual 3D models for transfer learning
the convolutional network (CNN) with adequate software can be seen in the Fig. 2.
Fig. 2. Detailed schematic of the CNN process using input data from 3D models.
2.3 Input Images for Teaching

Input samples are generated as PNG images, for each of the components were gen-
erated 36 frames with different materials and an adequate background:
• chrome material, wood table base,
• brass material, base steel plate,
• steel material, polished rock workplace.
Material texture is combined with suitable background that justify of visibility of
teaching object, currently this process is handled manually. Background mate-rial was
selected with condition of maximize brightness range (high brightness - wood, middle
brightness - steel, low brightness - rock). This selection of backgrounds can provide
minimal sensitivity of trained CNN model to changes in environments light condition.
Teaching samples use proportional dimensions for M12 screw with adequate nut and
washer size. Two sets of samples were created for teaching process, the first bigger set
is used for learning and second one was used for evaluation of teaching progress. In
teaching process currently missing samples with overlapping, because this was our first
experiment which should to confirm that virtual samples are suitable for convolutional
model teaching and it can reach comparable results like photo input samples from
produced parts.
Generation process of input samples uses very simple rendering engine (Cycles)
which is a part of Blender software. Default internal rendering engine of Blender
wasn’t suitable for rendering industrial parts (shiny metal surfaces).
All images are generated with same resolution 960x540 which is enough to train
part details by CNN. Generated input images are shown in the Fig. 3.
Fig. 3. Example set from teaching samples (nut, screw, washer).
XML file contains these basic data: image size, part type and object location within
the scene.
The graphical generation process with explanation of XML generation from 2D

image is shown in the Fig. 4.
Fig. 4. Blender 2D simulated image, part position generated by OpenCV, XML structure.
The CNN model Faster RCNN Inception v2 was selected for testing because of
high accuracy, but with disadvantage of network size and recognition speed. The CNN
model Mobilenet V2 is more suitable for Smart glasses (Integrated Android OS), but
this is very limited to input image size, because it is optimized for lower resolution,
usually with image resolution 300 300. This lower resolution not provide reliable
results in multiple detection of object (nut, screw, washer) in one picture, because
object lose details for example in screw threads and edges. Mobilenet V1 or V2
provides much more performance and two- or three-times smaller network configu-
ration file with fixed weights. Inception model occupies about 52 MB size on HDD in
opposite to MobileNet V2 which occupies only about 19 MB. For teaching process, we
doesn’t need teaching from scratch. We used learned model with general dataset pets
and coco with almost the same results. Teaching process only remove last layers which
we need to teach again by transfer learning.
2.4 Hardware Implementation to Soc and AR Smart Glasses

For assisted assembly we need transform trained model from PC to wearable device,
for example an embedded system or VR/AR device. Other possible developments are
augmented devices with SLAM algorithm [15], in combination with AR-based worker
support system for human-robot collaboration [16], an anchoring support system with
AR framework [17].
Our development is aimed for assisted assembly of product with augmented reality
to check, help and improve montage process. Deployment principle in flowchart form
for software/hardware implementation is shown in the Fig. 5.
Fig. 5. Deployment solution to real devices.

We chose as mobile testing platform embedded system ODROID C2 with the

ARM64 processor and OpenCV library (C++ version) installed to test the learned DNN
network (Fig. 6, left). Results of recognition precision and delays can be acquired by
internal web server or remote command line SSH shell from PC [18].
For real implementation to assisted assembly, it is very convenient to import a
solution to specialized device with augmented reality. We choose Epson Moverio
BT350 with transparent display and Android 5.1 OS (Fig. 6, right).
Fig. 6. ODROID with Linux OS, OpenCV test with command line application, SMART
Glasses for AR Epson Moverio BT350.
Development of software combines Unity environment for visualization and

OpenCV C++ for DNN models processing.
Hardware implementation require combination of three software:
• TensorFlow (Python) – Teaching CNN,
• Android Studio (Java/C++) – CNN execution,
• Unity (C#) – 3D engine – Data visualization.
Transform learning process was tested in TensorFlow framework and scripting
language Python with Anaconda environment to separate CNN teaching model.
Experimental transfer learning was speedup by CUDA cores by NVIDIA GTX1060
graphics card. The first evaluation of results was tested in python on PC platform.
Android studio is used for programming unique DLL plugin for fast CNN model
execution written in C++ language and uses OpenCV frameworks. This DLL is
developed and maintain in our department. The plugin is currently compiled for x64,
armv8 and armv7 Soc.
Unity framework provides very simple interface for creating simple 3D scene
compatible with Android, Linux and Windows, etc. Control logic is written in C#
language and provides image input to DLL plugin. A little speedup we can acquire by
translating C# to C++ by IL2CPP compiler inside of Unity build system.
Embedded device SoC (Odroid C2) with Armbian OS provides much more per-
formance of DNN model execution time about 600 ms. Smart Glasses (Epson Moverio
BT350) with Android can run with delay about 1400 ms (Intel Atom).
For comparison, we execute CNN model on standard PC with Intel i5-8400 pro-
cessor, which reach usable delay of 43 ms. The main problem of slow CNN model
execution on embedded platform are probably slower ARM/Atom processors.
Learning process was conducted for industry standardized parts (nuts, bolts, washers).
We performed transform learning with 200,000 cycles with achieved precision of
mAP 1.8e−3 for positioning and 1.37e−3 for classification. The transform learning
process took 10 h and 33 min. Separate graphs of learning classification and local-
ization are shown in the Fig. 7.
The experimental solution (CNN model) was first tested with different virtual
Fig. 7. The learning process for classification of objects (left), position of the object (right)
samples with changed angles and background randomly to verify that learned model is
independent on material/background and recognition works universally in any condi-
tion. Final experiments were provided with real image of industrial parts as is shown in
Table 1.
For assisted assembly we need to know position of detected part because in field of
view (FOV) can be more recognized parts placed. We used teaching process with
single shot detection (SSD) for part location. Teaching procedure improved both
parameters in parallel. With selected range of backgrounds teaching process for
position reach stable improvements. Precision of part type classification teaching
process is much more unstable, probably due to small size of part in training sample
image.
The next experiments were provided with multiple detection of real parts in image
(nut, screw, washer) as it is shown in the Fig. 8. For multiple recognition the results are
very precise, we achieved mAP in range from 91% to 99%.
The preparation of 3D virtual samples was not with part overlays, learned CNN
model doesn’t provide this feature for experiments. Preparation of part overlays is very
complicated for implementation to automated scripting cycle in Blender and currently it
must be prepared manually. Our idea for automated 3D models parts overlay genera-
tion is implementation of basic Newnton dynamics with free fall of parts to fixed table.
This principle can create natural overlay with all tested parts and without body
collision.
Table 1. Reliability verification of learned CNN model by using real samples.
Part Training: example of virtual part samples Testing: mAP

Name (changed material/background) Real samples
0.96
Screw
/
M12
0.99
0.99
Nut
/
M12
0.93
0.99
Washer
/
12
0.99
Fig. 8. Multiple recognition for all parts in one raster image
4 Conclusion
It is clear from the results of the experiments with 3D virtual model training technics
that CNN can be learned without real frames. We get probability from 0.93 to 0.99 in
individual recognition and 0.91 to 0.99 in multiple object recognition.
The reached results show that virtual models for teaching CNN are a perspective
way with significant efficiency. The first implementation is proposed to augmented
devices used in assisted assembly process.
Next software improvement will be web interface for automated sample generation
in user friendly interface. User upload 3D model only from design software by web
interface and cloud will teach CNN model and upload it to Android device hardware.
Further experimental work will be focused on learning and testing other pretrained
models, mainly minimized CNN models (MobileNet V2 or V3 with SSD) by the same
methodology with virtual 3D models, which are much faster on Android mobile
platform. Very interesting part for next experiments can be comparison of execution
time of the same CNN model in different CNN framework: TensorFlow, Caffe, Torch,
Microsoft Cognitive Toolkit, MXNet.
The next speed improving can be realized by transformation model from basic
embedded platform (ODROID Soc) to specific board with CUDA graphical processing
unit (GPU) for example NVIDIA Tegra TX1. Android implementation (Epson BT-350
AR) can significantly speedup by execution on hardware with APU, for example
Huawei P20 Pro mobile phone with combination of Google Cardboard VR mobile
holder. The maximal performance of object recognition speed with trained CNN model
is possible only on PC platform with external graphics card Nvidia 10x0 or AMD Vega
series with VR headset HTC Vive Pro.
Acknowledgments. This work was supported by the Slovak Research and Development
Agency under the contract No. APVV-15-0602 and also by the Project of the Structural Funds of
the EU, ITMS code: 26220220103.
References
1. Židek, K., Maxim, V., Piteľ, J., Hošovský, A.: Embedded vision equipment of industrial
robot for inline detection of product errors by clustering-classification algorithms. Int.
J. Adv. Rob. Syst. 13(5), 1–10 (2016)
2. Židek, K., Piteľ, J., Hošovský, A.: Machine learning algorithms implementation into
embedded systems with web application user interface. In: Proceedings of the IEEE 21st
International Conference on Intelligent Engineering Systems 2017 (INES 2017), pp. 77–81.
IEEE (2017)
3. Pavlenko, I., Trojanowska, J., Ivanov, V., Liaposhchenko, O.: Scientific and methodological
approach for the identification of mathematical models of mechanical systems by using
artificial neural networks. Lecture Notes in Electrical Engineering, pp. 299–306. Springer,
Cham (2019)
4. Pavlenko, I., Simonovskiy, V., Ivanov, V., Zajac, J., Pitel, J.: Application of artificial neural
network for identification of bearing stiffness characteristics in rotor dynamics analysis.
Lecture Notes in Mechanical Engineering, pp. 325–335. Springer, Cham (2019)
5. Židek, K., Hosovsky, A., Piteľ, J., Bednár, S.: Recognition of assembly parts by
convolutional neural networks. Lecture Notes in Mechanical Engineering, pp. 281–289.
Springer, Cham (2018)
6. Sarkar, K., Varanasi, K., Stricker, D.: Trained 3D models for CNN based object recognition.
In: Proceedings of the 12th International Joint Conference on Computer Vision, Imaging and
Computer Graphics Theory and Applications (Visigrapp 2017), vol 5, pp. 130–137 (2017)
7. Socher, R., Huval, B., Bhat, B., Manning, C.D., Ng, A.Y.: Convolutional-recursive deep
learning for 3D object classification. In: NIPS (2012)
8. Su, H., Qi, C.R., Li, Y., Guibas, L.J.: Render for CNN: viewpoint estimation in images using
CNNs trained with rendered 3D model views. In: Proceedings of the IEEE International
Conference on Computer Vision (2015)
9. Tian, Y., Li, X., Wang, K., Wang, F.Y.: Training and testing object detectors with virtual
images. IEEE/CAA J. Autom. Sin. 5, 539–546 (2018)
10. Nie, W.Z., Cao, Q., Liu, A.A., Su, Y.T.: Convolutional deep learning for 3D object retrieval.
Multimed. Syst. 23, 325–332 (2017)
11. Peng, X.C., Sun, B.C., Ali, K., Saenko, K.: Learning deep object detectors from 3D models.
In: IEEE International Conference on Computer Vision, pp. 1278–1286 (2015)
12. Karwasz, A., Trojanowska, J.: Using CAD 3D system in ecodesign - case study. In:
Golinska-Dawson, P., Kolinski, A. (eds.) Efficiency in Sustainable Supply Chain.
EcoProduction (Environmental Issues in Logistics and Manufacturing), pp. 137–160.
Springer, Cham (2017)
13. Nguyen, C.H.P., Choi, Y.: Triangular mesh and boundary representation combined approach
for 3D CAD lightweight representation for collaborative product development. J. Comput.
Inf. Sci. Eng. 19, 9–11 (2018)
14. Cucos, M.M., Pista, I.M., Ripanu, M.I.: Product engineering design enhancing by
parameterizing the 3D solid model. In: MATEC Web Conference, vol. 178, p. 05011 (2018)
15. Gao, Q.H., Wan, T.R., Tang, W., Chen, L., Zhang, K.B.: An improved augmented reality
registration method based on Visual SLAM. In: International Conference on Technologies
for E-Learning and Digital Entertainment, pp. 11–19. Springer, Cham (2017)
16. Liu, H., Wang, L.: An AR-based worker support system for human-robot collaboration.
Procedia Manuf. 11, 22–30 (2017)
17. Takaseki, R., Nagashima, R., Kashima, H., Okazaki, T.: Development of anchoring support
system using with AR toolkit. In: 2015 7th International Conference on Emerging Trends in
Engineering & Technology (ICETET), pp. 123–127. IEEE (2015)
18. Židek, K., Janáčová, D., Hošovský, A., Piteľ, J., Lazorík, P.: Data optimization for
communication between wireless IoT devices and cloud platforms in production process. In:
3rd EAI International Conference on Management of Manufacturing Systems (MMS 2018),
pp. 1–8. EAI (2018)
3D Printed Hand Exoskeleton - Own Concept
Jakub Kopowski1, Izabela Rojek2(&), Dariusz Mikołajewski2,

and Marek Macko2
1
Institute of Psychology, Kazimierz Wielki University,
Chodkiewicza 30, 85-064 Bydgoszcz, Poland
2
Institute of Mechanics and Applied Computer Science,
Kazimierz Wielki University, Chodkiewicza 30, 85-064 Bydgoszcz, Poland
izarojek@ukw.edu.pl
Abstract. Numerous hand exoskeletons have been proposed in the literature

with the aim of assisting or rehabilitating results of neurodegenerative changes
due to ageing, stroke, traumatic brain disorder, spinal cord injury, or other
causes of hand disorders. Key issue becomes: how to investigate, improve, and
observe the effect of rehabilitation therapy. Thus many studies have been con-
ducted on evaluating the motor function quantitatively by developing various
types of robotic systems. Even though the robotic systems have been developed,
this task still constitutes challenge, thus functional evaluation of the hand has
been rarely investigated, because it is difficult to install a number of actuators or
sensors to the hand due to limited space around the fingers. Our article aims at
presentation of the own concept of 3D printed hand exoskeleton, associated
occupations and limitations.
Keywords: 3D printing Rapid prototyping Biodegradable materials

Biocompatibility Hand exoskeleton
1 Introduction
Numerous hand exoskeletons have been proposed in the literature with the aim of
assisting or rehabilitating results of neurodegenerative changes due to ageing, stroke,
traumatic brain disorder, spinal cord injury, or other causes of hand disorders. To
investigate, improve, and observe the effect of rehabilitation therapy, many Key issue
becomes: how to investigate, improve, and observe the effect of rehabilitation therapy.
Thus many studies have been conducted on evaluating the motor function quantita-
tively by developing various types of robotic systems. Even though the robotic systems
have been developed, this task still constitutes challenge, thus functional evaluation of
the hand has been rarely investigated, because it is difficult to install a number of
actuators or sensors to the hand due to limited space around the fingers [1, 2].
Our article aims at presentation of the own concept of 3D printed hand exoskeleton,
associated occupations and limitations.

https://doi.org/10.1007/978-3-030-18715-6_25
3D Printed Hand Exoskeleton - Own Concept 299
2 3D Printing Unique Properties
3D printing and associated 3D technologies (3D scanning, computer-aided design –

CAD) is technology, which becomes inspiration for new generation of scientists,
inventors, designers, and specialists in applications of such novel technologies in
particular areas, such as medicine. Knowledge and experience concerning this specific
area of exploitation is worth for focus on due to unique opportunities and possible
weaknesses.
2.1 Technology Itself

Development of low-cost printing can significantly increase accessibility of medical
devices such as exoskeletons without worsening their quality. Early diagnosis and
skilled therapeutic process, including rehabilitation and assistive technology such as the
exoskeleton, can stop the progress of a disease, and even cause recovery within
activities of daily living. Described invention is cheap, patient-tailored, available for
children, develops with age and health status, and supports movement in all joints of
the hand.
3D printing constitutes technique of fabricating physical models from a 3D digital
image. The ultimate goal is to achieve the development of functional abilities of the
patients within medical and health sciences, including human organs and tissues, to
overcome limitations of surgical planning, staff and patient education, rehabilitation,
organ transplantation created by the neurodegenerative disorders, lack of organ donors
and life-long immunosuppression. Aforementioned technologies may provide precision
medicine approach to assistive technology and human organ fabrication using 3D
scanning, 3D printed technology, and reverse engineering. Digital volumetric data are
then usually collected by imaging of a patient, i.e. 3D scan, CT or MRI images (thanks
to volume rendering and medical image reconstruction), etc. Aforementioned proce-
dure is followed by computational modeling based on created digital 3D image. Even
very complex structure, unable to produce in the other way, can be created using a 3D
printer. There are many potential patient-specific applications of 3D printing in med-
icine, useful in every day clinical practice: from bandages and assistive technology
devices through individually printed (plastic, metal) implants and synthetic electronic
devices to tissues providing innervation. Exact copy of the complex shape (in the
bioprinting also: organization of natural tissue) required to more effective support,
repair or replace damaged tissues or organs. However, aforementioned technique still
needs to be optimized. Moreover, more knowledge and experience is required [3, 4].
Current open source packages and a low-cost desktop 3D printers allow to convert
various modalities of medical images to digital resources, which are useful to enhance
our understanding of the geometric structure and complex spatial nature of anatomical
organs and its co-operation, including biomechanical understanding of the limb
movement and anatomical relationship between organs and complicated surrounding
structures. Secondary task is the verification of the accuracy of reconstruction, both
hard organs (bones, etc.) and soft tissue organs (skin, muscles, blood vessels, etc.). The
accuracy of the printed product is limited by the quality of organ models, 3D scanning,
and 3D printing. Sometimes exactness of printing technique and features of materials
300 J. Kopowski et al.
available constitutes key issue [5]. 3D models allow interaction with real objects such
as implantable electrode grids despite using desktop 3D printers producing relatively
inexpensive, good quality. To sum up: the applications and limitations of 3D printing
thanks to its features may revolutionize the production process of medical devices, but
despite demonstrated efficacy is still in the beginning of its development [3–14].
2.2 Biodegradable Materials

Various materials for 3D printing may be taken into consideration. Traditional 3D
printing methods involve the use of various materials, including heat, toxic solvents or
photoinitiators used e.g. for fabrication of synthetic scaffolds. Biodegradable materials
undergoing degradation in vivo are common employed e.g. to manufacture tissue
engineering scaffolds. Water-based 3D printing for synthetic materials may be another
breakthrough in the aforementioned area [15].
2.3 Biocompatibility
Biocompatibility of the fabricated 3D printed parts should be studied, including acute
toxicity, pyrogenic reaction, etc. Skin and body fluids contact may cause unexpected
changes in the structure of 3D printed porous material. Thus biomechanical properties
and high biocompatibility have similar importance. Emphasizing aforementioned fea-
ture has a good application prospect in the prototyping of many 3D printed parts,
including not only exoskeletons, but also tissue-engineered bone scaffolds [16].
3 Hand Exoskeleton - Own Concept
Invention described here is the result of latest scientific research. It is based on the
novel knowledge in the fields of biomedical engineering, experimental biomechanics,
material engineering, rehabilitation and physiotherapy.
Construction of the prototype and current research is conducted in the Institute of
Mechanics and Applied Computer Science, Kazimierz Wielki University in Bydgoszcz
(Poland). Interdisciplinary team of scientists headed by Professor Izabela Rojek con-
sists of experts in technical sciences: biocybernetics and biomedical engineering,
experimental biomechanics, mechatronics, and artificial intelligence applications. Such
approach allows for comprehensive studies of various designs, evaluating their effi-
ciency, and applying the best actually available solutions derived from disciplines
mentioned above.
Relatively quick work on aforementioned solution increases chances for rapid
development of prototypes ready for testing in homes on a large scale. Next stage of the
research beginning in December 2019 covers development of the whole family of the
prototypes – the most promising of them will be chosen for further studies.
Participation of well-established research team should help in achieving both sci-
entific and commercial success both in preventive medical devices segment of the
market.
3.1 General Assumptions

An exoskeleton is a distinctive kind of robot to be worn as an overall or frame,
effectively supporting, or in some cases substituting for, the user’s own movements. An
exoskeleton is a technical tool that expands and improves selected abilities of the user.
From a medical point of view, it can serve as a multi-purpose medical device increasing
patient’s possibilities in a way closer to natural human mobility and as an option in
restoring upper-limb functions. Moreover, exoskeletons are perceived as probably
more effective than the traditional assistance and support of therapists and rehabilitative
device.
Exoskeletons are controlled by the user’s movements and do not need any external
control terminal (with the exception of a service terminal).
The main parts are usually: the frame; the power system, including engines,
actuators and batteries; and the control system with sensors. Many subtypes of passive
and active exoskeletons are recognized.
Stages of the work are following:
– detailed description of the hand anatomy and biomechanics, including experimental
studies,
– computational model of the healthy hand,
– families of computational models of the hand with various disabilities, mainly based
on neurodegenerative disorders,
– construction of the hand exoskeleton based on own project,
– creating digital template of the exoskeleton taking into consideration shape and
dimensions of the hands (left, right), handedness, kind and level of the deficit, and
the other particular features remarkable for the individual patient,
– creating real exoskeleton for the particular real patients (Fig. 1),
– testing efficacy of the aforementioned exoskeleton.
Further research are going to achieve following goals:
– clear procedure of hand(s) assessment, scanning, and imaging toward ability of the
hand exoskeleton use,
– clear procedure of hand exoskeleton design based on pre-programmed patterns,
– objective assessment of efficacy of the hand exoskeleton in homogenous group of
patients (in the co-operation to the Department of Physiotherapy, Ludwik Rydygier
Collegium Medicum in Bydgoszcz, Nicolaus Copernicus University in Toruń,
Poland),
– based on the results of the aforementioned studies: development of the whole family
of solutions dedicated to particular homogenous groups of patients,
– increasing of the technology readiness level (TRL) of the aforementioned solutions.
Current studies are focused twofold on:
– computational model of the hand based on both traditional scans, medical imaging,
and dedicated solutions (e.g. for gesture control purposes),
– computational models of the hand exoskeleton using Design Thinking paradigm,
ready-to-use in 3D printing solutions.
Studies on control systems allow for certain adaptation of the current technologies
to the novel abilities. But we should be aware that hands are very important and basic
tools of people. Hand movements are the most precise, challenging in modelling and
replication, and every tool should be specially pre-prepared, according to the possi-
bilities and needs of the particular user, both healthy and with hand disorders.
3.2 Strengths and Occupations

Current solutions of hand exoskeletons are mainly early prototypes with low TRL.
They are often made according to technologies used 10 years before, moreover they
don’t use reverse engineering and don’t support all joints of the hand, including wrist.
They haven’t prove their usability in mass production so far.
Computer Aided Design (CAD)

Real handí models
Individual fitting of hand exoskeleton
features (including deformities,
amputations, etc.)
Virtual 3D 3D printing
model
Testing of the
3D scanning physical models,
Analysis, testing, modernization, fast
assessment, prototyping,
Scanning modernization,
of the real hand preparation of files
adaptation, need for printing
personalization of the end-products
of the particular
solution
Other 3D technologies
Gathering of data need for models used

in the case of defects or deformities
of bones and tissues
Fig. 1. Idea of the hand exoskeleton production process.

The invention solves a problem of simple, low-cost, quickly produced, exact,

patient-tailored, easy-to-use hand exoskeletons for children and adults thanks to the
comprehensive application of 3D scanning, 3D printing, and reverse engineering.
Novelties important for exoskeleton users are:
– support of all upper limb joints,
– support of the movement,
– lack of limitations due to the user’s dimensions of limbs,
– decreased time and effort needed for fitting and pre-setting the exoskeleton before
its first use,
– possibility of continuous exoskeleton fitting thanks to adjustable parts,
– convenience and comfort in all-day use,
– possibility of transporting.
3.3 Weaknesses and Limitations

Exoskeletons are still at the early stages of their development. They need detailed tech-
nical and clinical research not only in the area of safety, but also in terms of their influence
on the human body, biomechanics and mind. It seems that in the future exoskeletons may
become a form of therapy in neurological diseases and neurorehabiltiation.
4 Discussion and Conclusions
4.1 Other Studies

Due to rapid development of 3D printing and mechatronic technology and subsequent
recent breakthrough we selected for compartmental study only the most recent publi-
cations. Burns and al. proposed a 3D printed soft hand exoskeleton HEXOES (Hand
Exoskeleton with Embedded Synergies) supporting metacarpophalangeal (MCP) and
proximal interphalangeal/interphalangeal (PIP/IP) joints. Observed maximum flexion
angles were of 53.7 ± 16.9° and 39.9 ± 13.4°, respectively; and maximum MCP and
PIP angular velocities of 94.5 ± 41.9°/s and 74.6 ± 67.3°/s, respectively [16]. In the
study by Kim et al., a hand exoskeleton was developed to satisfy required hand
functions including spasticity of finger flexors, finger independence, and multi-digit
synergy. The hand exoskeleton was composed with the four 4-bar linkages, two
motors, and three loadcells (for each finger). It was able to flex/extend the metacarpal
(MCP) and proximal interphalangeal (PIP) joints independently [17]. Jo et al. proposed
design of a wearable hand exoskeleton system for exercising flexion/extension of the
fingers (1 degree of freedom - DOF).
A prototype consisted of two parts:
– one part for the thumb,
– the other for rest fingers.
To customize the hand exoskeleton for the user:

– polynomial curve was adjusted to the joint range of motion (ROM) of the user,
– optimal design of the exoskeleton structure was obtained using the optimization
algorithm [18].
4.2 Limitations of Own Studies

Ultimate solution needs for standardization of procedures, data bases and tools for
interoperability purposes. Free access to medical data sets, even limited to staff and
students, needs for obeying ethical and legal rules.
4.3 Directions for Further Research

Directions for further studies:
– coupling with the control loop as a test-bed for synergy-based clinical hand
rehabilitation,
– advanced control systems based on brain-computer interface (BCI) technology [19–
21],
– novel rehabilitation protocols incorporating hand exoskeleton [22],
– novel solutions based on computational intelligence [23–31],
– application of a new type of composites, in which matrix is manufactured addi-
tively, whereas infill is a polyurethane resin. The process of manufacturing these
composites is invented and patented by authors [31–34],
– as a step further: 3D printing of combination of cells, growth factors, biomaterials
and even supporting plastic parts and electronic devices in the field of regenerative
medicine.
Further technical developments are needed to solve current limitations and problems
with exoskeletons: lighter and stronger materials for the frame, more powerful actuators
and long life batteries, more sophisticated control systems, and more efficient processes
of fitting and customizing, which requires severe changes in the area of young specialists
education and patients’ provision in assistive technology devices [1, 2].
4.4 Conclusions
Proposed 3D printed hand exoskeleton may create another breakthrough in the func-
tional neurorehabilitation of patients with various kinds and levels hand disorders, both
as temporary (e.g. during recovery, including people with weakness within upper limb)
and permanent solution.
Current results and estimates indicate that the mechanical design enhanced using
computational intelligence has ability to functionally cover range of motion and
angular velocity characteristics that meet the requirements for synergy-based control.
References
1. Mikołajewska, E., Mikołajewski, D.: Exoskeletons in neurological diseases-current and
potential future applications. Adv. Clin. Exp. Med. 20(2), 227–233 (2011)
2. Mikołajczyk, T., Kłodowski, A., Mikołajewska, E., Walkowiak, P., Berjano, P., Villafañe, J.
H., Aggogeri, F., Borboni, A., Fausti, D., Petrogalli, G.: Design and control of system for
elbow rehabilitation: preliminary findings. Adv. Clin. Exp. Med. (2018). https://doi.org/10.
17219/acem/74556
3. Radenkovic, D., Solouk, A., Seifalian, A.: Personalized development of human organs using
3D printing technology. Med. Hypotheses 87, 30–33 (2016)
4. Visser, J., Melchels, F.P., Dhert, W.J., Malda, J.: Tissue printing; the potential application of
3D printing in medicine. Ned. Tijdschr. Geneeskd. 157(52), A7043 (2013)
5. Naftulin, J.S., Kimchi, E.Y., Cash, S.S.: Streamlined, inexpensive 3D printing of the brain
and skull. PLoS One 10(8), e0136198 (2015)
6. Gür, Y.: Additive manufacturing of anatomical models from computed tomography scan
data. Mol. Cell Biomech. 11(4), 249–258 (2014)
7. Mikołajewska, E., Macko, M., Mikołajewski, D., Ziarnecki, Ł., Stańczak, S., Kawalec, P.:
Medical and military applications of 3D printing. Zeszyty Naukowe Wyższej Szkoły
Oficerska Wojsk Lądowych im. generała Tadeusza Kościuszki 179(1), 128–141 (2016)
8. Macko, M., Mikołajewska, E., Szczepański, Z., Augustyńska, B., Mikołajewski, D.:
Repository of images for reverse engineering and medical simulation purposes. Med. Biol.
Sci. 30(3), 23–29 (2016)
9. Macko, M., Szczepański, Z., Mikołajewski, D., Mikołajewska, E., Nowak, J., Listopadzki,
S.: The method of artificial organs fabrication based on reverse engineering in medicine. In:
III International Scientific Conference: Morpho-Biomechanical and Psycho-Physical Aspects
of Youth Lifestyle in V4 Countries, vol. 45. Institute of Physical Education, Kazimierz
Wielki University in Bydgoszcz (2016)
10. Mikołajewska, E., Macko, M., Ziarnecki, Ł., Stańczak, S., Kawalec, P., Mikołajewski, D.:
3D printing technologies in rehabilitation engineering. J. Health Sci. 4(12), 78–83 (2014)
11. He, Y., Xue, G.H., Fu, J.Z.: Fabrication of low cost soft tissue prostheses with the desktop
3D printer. Sci Report, vol. 4, p. 6973 (2014)
12. Choonara, Y.E., du Toit, L.C., Kumar, P., Kondiah, P.P., Pillay, V.: 3D-printing and the
effect on medical costs: a new era? Expert Rev. Pharmacoecon. Outcomes Res. 16(1), 23–32
(2016)
13. Gu, Q., Hao, J., Lu, Y., Wang, L., Wallace, G.G., Zhou, Q.: Three-dimensional bio-printing.
Sci. China Life Sci. 58(5), 411–419 (2015)
14. Schubert, C., van Langeveld, M.C., Donoso, L.A.: Innovations in 3D printing: a 3D
overview from optics to organs. Br. J. Ophthalmol. 98(2), 159–161 (2014)
15. Hung, K.C., Tseng, C.S., Hsu, S.H.: Synthesis and 3D printing of biodegradable
polyurethane elastomer by a water-based process for cartilage tissue engineering applica-
tions. Adv. Healthc. Mater. 3(10), 1578–1587 (2014)
16. Burns, M.K., Van Orden, K., Patel, V., Vinjamuri, R.: Towards a wearable hand exoskeleton
with embedded synergies. In: Conference on Proceedings of the IEEE Engineering in
Medicine and Biology Society, pp. 213–216 (2017)
17. Kim, S., Lee, J., Park, W., Bae, J.: Quantitative evaluation of hand functions using a
wearable hand exoskeleton system. In: IEEE International Conference on Rehabilitation
Robotics, pp. 1488–1493 (2017)
18. Jo, I., Lee, J., Park, Y., Bae, J.: Design of a wearable hand exoskeleton for exercising
flexion/extension of the fingers. In: IEEE International Conference on Rehabilitation
Robotics, pp. 1615–1620 (2017)
19. Chowdhury, A., Meena, Y.K., Raza, H., Bhushan, B., Uttam, A.K., Pandey, N., Hashmi, A.
A., Bajpai, A., Dutta, A., Prasad, G.: Active physical practice followed by mental practice
using BCI-driven hand exoskeleton: a pilot trial for clinical effectiveness and usability.
IEEE J. Biomed. Health Inform. 22(6), 1786–1795 (2018)
20. Wierzgała, P., Zapała, D., Wójcik, G.M., Masiak, J.: Most popular signal processing
methods in motor-imagery BCI: a review and meta-analysis. Front. Neuroinform. 12, 78
(2018)
21. Wójcik, G.M.: Self-organising criticality in the simulated models of the rat cortical
microcircuits. Neurocomputing 79, 61–67 (2012)
22. Frolov, A.A., Mokienko, O., Lyukmanov, R., Biryukova, E., Kotov, S., Turbina, L.,
Nadareyshvily, G., Bushkova, Y.: Post-stroke rehabilitation training with a motor-imagery-
based brain-computer interface (BCI)-controlled hand exoskeleton, a randomized controlled
multicenter trial. Front. Neurosci. 11, 400 (2017)
23. Rojek, I.: Neural networks as performance improvement models in intelligent CAPP
systems. Control Cybern. 39(1), 55–68 (2010)
24. Rojek, I.: Technological process planning by the use of neural networks. Artif. Intell. Eng.
Des. Anal. Manuf. 31(1), 1–15 (2017)
25. Rojek, I.: Neural networks as prediction models for water intake in water supply system. In:
Lecture Notes in Artificial Intelligence, vol. 5097, pp. 1109–1119. Springer (2008)
26. Rojek, I.: Hybrid neural networks as prediction models. In: Lecture Notes in Artificial
Intelligence, part II, vol. 6114, pp. 88–95. Springer (2010)
27. Dostatni, E.: Recycling-oriented eco-design methodology based on decentralised artificial
intelligence. Manag. Prod. Eng. Rev. 9(3), 79–89 (2018)
28. Dostatni, E., Diakun, J., Grajewski, D., Wichniarek, R., Karwasz, A.: Multi-agent system to
support decision-making process in design for recycling. Soft Comput. 20(11), 4347–4361
(2016)
29. Macko, M., Flizikowski, J., Szczepański, Z., Tyszczuk, K., Śmigielski, G., Mroziński, A.,
Czerniak, J., Tomporowski, A.: CAD/CAE applications in Mill’s design and investigation.
In: Computer Aided Engineering. Lecture Notes in Mechanical Engineering, pp. 343–351.
Springer (2017)
30. Czerniak, J., Macko, M., Ewald, D.: The CutMAG as a new hybrid method for multi-edge
grinder design optimisation. In: Advances in Intelligent Systems and Computing, vol. 401,
pp. 327–337. Springer (2016)
31. Czerniak, J., Ewald, D., Macko, M., Śmigelski, G., Tyszczuk, K.: Approach to the
monitoring of energy consumption in eco-grinder based on ABC optimization. In:
Communications in Computer and Information Science, vol. 521, pp. 516–529 (2015)
32. Górski, F., Wichniarek, R., Zawadzki, P., Hamrol, A.: Computation of mechanical properties
of parts manufactured by fused deposition modeling using finite element method. In:
Advances in Intelligent Systems and Computing, vol. 368, pp. 403–413. Springer (2015)
33. Górski, F., Wichniarek, R., Kuczko, W., Zawadzki, P., Buń, P.: Strength of ABS parts
produced by fused deposition modeling technology - a critical orientation problem. Adv. Sci.
Technol. Res. J. 9, 12–19 (2015). https://doi.org/10.12913/22998624/2359
34. Górski, F., Kuczko, W., Wichniarek, R., Hamrol, A.: Mechanical properties of composite
parts manufactured in FDM technology. Rapid Prototyp. J. (2018). https://doi.org/10.1108/
RPJ-11-2016-0197
Mechanical Analysis of Leaf Springs Realized
by Additive Technologies
Ciprian I. Rizescu, Daniel Besnea, Dana Rizescu(&), Edgar Moraru,

and Victor Constantin
Department of Mechatronics and Precision Mechanics,

Politehnica University of Bucharest, Bucharest, Romania
dana.rizescu@upb.ro
Abstract. The paper deals with a new study for leaf springs. The authors
considered two materials: ABS (Acrylonitrile Butadiene Styrene) and PLA
(Polylactic Acid), and two fill factors: 10% and 20%. The springs have been
realized using FDM (Fused Deposition Modeling) technology and have four
different thicknesses (2.5, 3, 3.5, 4 mm). Three samples of each type were
manufactured with a total of 48 springs. In about 45 min, the leaf spring with
good aspect was realized. The main mechanical characteristics of the elastic
elements made by additive technologies were determined experimentally,
comparing them with those calculated theoretically, obtaining satisfactory
results. The spring model behavior was analyzed using Inventor 2016 envi-
ronment. Two parameters were considered: displacement and stress. Interesting
conclusions about the stiffness of the springs have been deduced.
Keywords: Additive technologies Fused deposition modeling Leaf springs

Thermoplastics
1 Introduction
In the construction of many mechanisms, and especially in the construction of mea-

suring and control devices, there are implemented pre-formed actuating systems. They
are used for both functional and gauge considerations. The constructive types of these
elastic elements, although very diverse, have a profile with common sections in geo-
metric aspect, consist of circle arches and straight segments of various lengths repre-
senting the sides of the spring. For the design of these springs, it is necessary to know
the mechanical characteristics of the material - bending resistance and the longitudinal
elastic modulus - as well as the size, direction and direction of force applied to the
elastic element in exploitation [1, 2]. U-shaped leaf spring with non-equal arms is
represented in the Fig. 1. In this case, due to its shape, the spring is divided into three
parts corresponding to the three elements: a straight portion of length u, a semicircle of
radius r and a straight portion of length v [3].
Force P acts angularly at the angle a, at the free end. Therefore, P breaks down on
the two Cartesian directions and results in the Px and Py components respectively
(Fig. 2).

https://doi.org/10.1007/978-3-030-18715-6_26
308 C. I. Rizescu et al.
Fig. 1. U shaped leaf spring
Fig. 2. Geometry of leaf spring
The springs have different thicknesses h = 2.5, 3.0, 3.5 and 4.0 mm. According to
these thicknesses there were determined the characteristics Force-Deflection presented
in the subsequent pictures. P1 corresponds to h = 2.5 mm, P2 corresponds to the
h = 3.0 mm, P3 corresponds to h = 3.5 mm and P4 corresponds to h = 4.0 mm. The
deflection range between 0 to 5 mm is for all springs. The deflection will be:
p P r3
f ¼ ð1Þ
2 EI
Where P [N] is a force in Newton, E [N/m2] – Young modulus of material, I [m4] -

moment of inertia. For the leaf spring, the moment of inertia will be: I = b h3/12.
The purpose of the work is to produce laminar springs with linear characteristics
that comply with Hooke’s law with use in various industrial and others areas.
Mechanical Analysis of Leaf Springs Realized by Additive Technologies 309
Garanger et al. [4] have also approached similar themes, but the other geometries of
the springs. Saleh et al. [5] highlight other additive manufacturing processes for real-
izing elastic elements.
Additive technologies through which the elements of this study have been developed
are processes with a spectacular global development. 3D printing or additive tech-
nologies involve the creation of layer-by-layer objects by using materials such as
plastic filaments (fused deposition modeling technology), resins (vat photopolymeri-
sation technology), plastic or metallic powders (selective laser deposition technology).
Using a power source, such as a laser or a heated extruder, the layers of these materials
are solidified to form the finished part. The benefits of 3D printing include form
freedom, multi-sector applications, precision, speed and ability to reduce costs and
weight in parts. There are many areas where additive technologies have found their
place: medicine [6–12], aeronautics [13–15], military [16] and many others [17–19].
Fig. 3. Fused deposition modeling process
Realizing a prototype or a finished product using a three-dimensional electronic

model via additive technologies [20–25] today is an affordable service that differs from
other methods of making physical samples with high speed, quality and reasonable
cost. FDM (Fused Deposition Modeling) - three-dimensional printing technology, in
which the construction of the object is due to the melting of the plastic thread, which is
fed to the work surface through an extruder (Fig. 3) [26]. The advantages of FDM
systems result from the fact that the manufacturing process does not produce much
material in the form of waste, it uses economically accessible materials and FDM
systems are easy to use and do not require special conditions for installation and
operation. Disadvantages are due to the lower quality of the processed surfaces, mainly
because of the scale effect, lower precision (0.1–0.2 mm) and relatively small sizes that
can be manufactured. As material in FDM 3D printers, special plastic threads are used
for this purpose. Such threads can be 1.75 mm in diameter, or 2.85 mm. A huge
amount of plastic is equipped to work with FDM 3D printers, however not every
printer supports the whole range of materials offered by the market. This factor depends
on the melting temperature of a particular material and the maximum heating tem-
perature of the nozzle of the device. The most common materials in this area are ABS
(Acrylonitrile butadiene styrene) and PLA (Polylactic acid), which were used in this
study. ABS is considered better in terms of mechanical and thermal properties, but PLA
is more environmentally friendly [27].
3 Experimental
The production cycle begins with the processing of a three-dimensional digital model.
Development of the 3D model is carried out in special programs, in our case the spring
model was designed in CATIA V5 [28]. The printed leaf springs are shown in Fig. 4.
Fig. 4. ABS (acrylonitrile butadiene styrene) and PLA (polylactic acid) leaf springs
The next step is to load the 3D model into the slicer program. At this stage, the
necessary 3D printing settings (speed, temperature, etc.) are selected, and supporting
structures are generated in the presence of overhanging elements in the model. Upon
completion of the preparatory part, the program generates a control code (G-code) for
the 3D printer based on the digital data (model) and the selected settings. The Repli-
cator G software was used for this purpose. The interface of the software is shown in
Fig. 5. Then the control code is transferred to the 3D printer. A USB flash drive or SD
card with recorded G-code (in advanced 3D printers it is also possible to transfer a file
over a wireless Wi-Fi network) connects to the device, in the menu of which the
required file is selected and printing starts. Upon completion of 3D printing, the fin-
ished product is detached from the 3D printer platform, cleared of supports and excess
plastic.
Fig. 5. Replicator G software interface
The equipment used was Wanhao Duplicator 4S [29] and the obtained springs are
represented in Fig. 6.
Fig. 6. Resulted 3D printed leaf springs: A – ABS spring; B – PLA spring.
After the completion of the process of fabrication, the springs were acted on Hans
Schmidt’s HV 500 N test equipment (Fig. 7), which consists of a deflection mea-
surement system, an Imada force transducer and the experimental piece.
Fig. 7. Leaf springs on Hans Schmidt’s HV 500 N test equipment
4 Results
In total 48 elastic elements (three samples of each type) were manufactured and tested
from different materials (ABS and PLA), fill factors (10 and 20%) and thicknesses (2.5,
3, 3.5 and 4 mm). Their weight ranged from 5 to 8 grams, with an average of 6 g
(Fig. 8). All realized springs are enumerated in the Table 1.
Fig. 8. Weighing of the spring

Table 1. Realized leaf springs

Material (fill factor %) Thickness [mm] Weight [g]
Sample 1 Sample 2 Sample 3
ABS (10%) 2.5 4 5 4
3 5 5 5
3.5 6 6 6
4 6 6 6
ABS (20%) 2.5 5 5 5
3 5 5 5
3.5 6 6 6
4 6 6 6
PLA (10%) 2.5 6 5 5
3 6 6 5
3.5 7 6 6
4 8 6 6
PLA (20%) 2.5 6 5 5
3 7 6 5
3.5 8 6 6
4 8 7 6
It has been observed that the mass of the polylactic acid springs is generally greater
than the mass of the acrylonitrile butadiene styrene elastic elements due to its higher
density.
Fig. 9. Comparison of leaf springs (thickness 4 mm) with 10% and 20% fill factors
Figure 9 shows the comparison of 4 mm thick springs with different fill factors and
materials. The experimental result of the ABS elastic element with 4 mm thickness and
20% fill factor was compared to the theoretical ones (Eq. 1), achieving close results
(Fig. 10). It is to be noted that the value for the elastic modulus for ABS is considered
to be E = 1100 N/mm2. It can be observed from the figures that the stiffness increases
with increasing fill factor. After analyzing the all cases, for the same material and same
thickness of material does not matter the fill factor. In addition, for the same thickness,
and same fill factor, the deformation forces of polylactic acid are higher than those of
acrylonitrile butadiene styrene.
Fig. 10. Comparison between experimental and theoretical results of ABS leaf spring (thickness
4 mm and 20% fill factor).
5 Simulation
The authors developed a simulation study of leaf spring model. The spring model
behavior was analyzed using Inventor 2016 environment. Two parameters were con-
sidered: displacement and stress. For each material and each thickness there were
simulated the behavior of leaf springs. In Fig. 11 there are shown the simulation results
of displacement and von Misses stress for acrylonitrile butadiene styrene spring with
3 mm thickness and 10% fill factor [30–32]. Numerical simulation was performed in
order to compare obtained experimental and theoretical results. Analyzing the values
obtained from the numerical simulation of the displacement, it can be observed that the
results of the three methods are close and the methodology of this paper is validated.
Fig. 11. Displacements (left) and stresses (right) for ABS leaf spring h = 3 mm, 10% fill factor
6 Conclusions
It was developed a study of mechanical behavior of U shaped leaf springs. The springs
were manufactured from two different materials – ABS and PLA, different thicknesses
(2.5, 3, 3.5, 4 mm) and different fill factors (10% and 20%). Per total it has been
realized 48 (three samples) leaf springs which were tested on experimental stand and
conclusions are following: the spring rigidity increase with the fill factor increasing; for
the same fill factor, we found that PLA rigidity is higher than ABS rigidity; the highest
rigidity is obtained for PLA with fill factor 20%. In general, satisfactory almost linear
characteristics have been obtained, especially in the case of polylactic acid springs. The
fill factor does not matter for the same material and thickness. For the same thickness
and the same fill factor the PLA forces are higher than ABS forces or we can say that
the deformation forces of PLA are higher than those of ABS for the same thickness and
fill factor. The authors developed a simulation analysis in order to determine the
mechanical behavior of the leaf springs. The spring model displacements computed
using Eq. (1) are very close to obtained displacements using Inventor simulation. There
were considered: displacement and von Misses stress for ABS and PLA. In addition,
experimental and theoretical results are close. Additive technologies through which the
elements of this study have been developed are processes with a spectacular global
development with multiple advantages. These obtained thermoplastic springs in a short
time by additive technologies with good mechanical properties can be a promising
solution for areas where metallic springs are not recommended – chemical industry,
water purification, radiography and other domains. In the future, other materials for
these springs will be studied: Nylon FX 256 Extrafill, CarbonFill and biocompatible
materials in order to increase mechanical properties of the springs and to expand the
domains of usage. In addition, other spring types and geometries will be studied. Major
interest also presents additive manufacturing processes that use other operating prin-
ciples. For example, vat photopolymerisation method – an additive technology which
uses a liquid raw material and it is possible to obtain parts with superior surface
characteristics compared to fused deposition modeling process.
References
1. Demian, T., Tudor, D., Curita, I., Nitu, C.: Fundamentals of precision engineering
instruments design, vol. 1. Printed by Editor Tehnica, Bucharest, Romania (1984). (in
Romanian)
2. Demian, T., Palade, D., Curita, I.: Elastic elements in precision engineering instruments.
Printed by Editor Tehnica, Bucharest, Romania (1994). (in Romanian)
3. Rizescu, C.I.: Instruments and mechanisms for Precision engineering, Part I. Printed by
Editor Printech, Bucharest, Romania (2013). (in Romanian)
4. Garanger, K., Khamvilai, T., Feron, E.: 3D printing of a leaf spring: a demonstration of
closed-loop control in additive manufacturing. In: 2018 IEEE Conference on Control
Technology and Applications, CCTA 2018, pp. 465–470 (2018)
5. Saleh, M.A., Ragab, A.E.: Ti-6Al-4V helical spring manufacturing via SLM: effect of
geometry on shear modulus. In: Proceedings of the International MultiConference of
Engineers and Computer Scientists, IMECS 2013, Hong Kong, 13–15 March 2013, vol. II
(2013)
6. Vithani, K., Goyanes, A., Jannin, V., Basit, A.W.: An overview of 3D printing technologies
for soft materials and potential opportunities for lipid-based drug delivery systems. Pharm.
Res. 36(1), 4 (2019)
7. Buonamici, F., Furferi, R., Governi, L., Lazzeri, S., McGreevy, K.S., Servi, M., Talanti, E.,
Uccheddu, F., Volpe, Y.: A practical methodology for computer-aided design of custom 3D
printable casts for wrist fractures. Vis. Comput. 1–16 (2019)
8. Mulford, J.S., Babazadeh, S., Mackay, N.: Three-dimensional printing in orthopaedic
surgery: review of current and future applications. ANZ J. Surg. 86, 648–653 (2016)
9. Carfagni, M., Furferi, R., Governi, L., Servi, M., Uccheddu, F., Volpe, Y., et al.: Fast and
low cost acquisition and reconstruction system for human hand-wrist-arm anatomy. Proc.
Manuf. 11, 1600–1608 (2007)
10. Krujatz, F., Lode, A., Seidel, J., Bley, T., Gelinsky, M., Steingroewer, J.: Additive Biotech-
Chances, challenges, and recent applications of additive manufacturing technologies in
biotechnology. New Biotechnol. 39, 222–231 (2017)
11. Revilla, M.L., Klemm, I.M., García-Arranz, J., Özcan, M.: 3D metal printing - additive
manufacturing technologies for frameworks of implant-borne fixed dental prosthesis. Eur.
J. Prosthodont. Restor. Dent. 25(3), 143–147 (2017)
12. Chudinova, E.A., Surmeneva, M.A., Timin, A.S., Karpov, T.E., Wittmar, A., Ulbricht, M.,
Ivanova, A., Loza, K., Prymak, O., Koptyug, A., Epple, M., Surmenev, R.A.: Adhesion,
proliferation, and osteogenic differentiation of human mesenchymal stem cells on additively
manufactured Ti6Al4V alloy scaffolds modified with calcium phosphate nanoparticles, 176,
130–139 (2018)
13. Dubois, A., Farhat, C., Abukhwejah, A.H., Shageer, H.M.: Parameterization framework for
the MDAO of wing structural layouts. AIAA J. 56(4), 1–12 (2018)
14. Mami, F., Reveret, J.-P., Fallaha, S., Margni, M.: Evaluating eco-efficiency of 3D printing in
the aeronautic industry: 3D printing in aeronautics through eco-efficiency. J. Ind. Ecol. 21
(31) (2017)
15. Atzeni, E., Iuliano, L., et al.: Additive manufacturing as a cost‐effective way to produce
metal parts. In: 6th International Conference on Advanced Research in Virtual and Physical
Prototyping, VR@P 2013, 1–5 October 2013, pp. 3–8. CRC/Balkema, Leiria (2013)
16. Fiske, M., Edmunson, J., Fikes, J., Johnston, M., Case, M.P.: 3D printing in space: a new
paradigm for America’s military. Mil. Eng. 110(715), 74–76 (2018)
17. Attaran, M.: 3D printing: enabling a new era of opportunities and challenges for
manufacturing. Int. J. Eng. Res. Gen. Sci. 4, 2320–2324 (2016)
18. Waheed, S., Cabot, J-M., Smejkal, P., Farajikhah, S., Sayyar, S., Innis, P.C., Beirne, S.,
Barnsley, G., Lewis, T.W., Breadmore, M.C., Paull, B.: 3D printing of abrasive, hard and
thermally conductive synthetic micro-diamond-polymer composite using low-cost fused
deposition modelling printer. ACS Appl. Mater. Interfaces (2019). Just Accepted Manuscript
19. Bressan, L.P., Adamo, C.B., Quero, R.F., Pereira de Jesus, D., Fracassi da Silva, J.A.:
Simple procedure to produce FDM-based 3D-printed microfluidic devices with integrated
PMMA optical window. Anal. Methods (2019). Accepted Manuscript
20. Moraru, E., Besnea, D., Dontu, O., Gheorghe, G.I., Constantin, V.: Applications of additive
technologies in realization of customized dental prostheses. Int. J. Mech. Appl. Mech. 3, 67–
71 (2018)
21. Besnea, D., Rizescu, D., Rizescu, C., Dinu, E., Constantin, V., Moraru, E.: Additive
technologies and materials for realization of elastic elements. Int. J. Mech. Appl. Mech. 3,
249–254 (2018)
22. Yuan, L., Ding, S., Wen, C.: Additive manufacturing technology for porous metal implant
applications and triple minimal surface structures: a review. Bioact. Mater. 4(1), 56–70
(2019)
23. Liu, J., Sun, L., Xu, W., et al.: Current advances and future perspectives of 3D printing
natural-derived biopolymers. Carbohydr. Polym. 207, 297–316 (2019)
24. Gardan, J.: Smart materials in additive manufacturing: state of the art and trends. Virtual
Phys. Prototyp. 14(1), 1–18 (2019)
25. Besnea, D., Dontu, O., Constantin, V., Spanu, A., Rizescu, C., Moraru, E.: Applications of
additive technologies in the food industry. Int. J. Mech. Appl. Mech. 3, 245–248 (2018)
26. Adel, M., Abdelaal, O., Gad, A., Nasr, A.B., Khalil, A.: Polishing of fused deposition
modeling products by hot air jet: evaluation of surface roughness. J. Mater. Process.
Technol. 251, 73–82 (2018)
27. Farbman, D., McCoy, C.D.: Materials testing of 3D printed ABS and PLA samples to guide
mechanical design. In: ASME 2016 11th International Manufacturing Science and
Engineering Conference (2016)
28. Ghionea, I.: Computer aided design in CATIA V5. Theoretic elements and applications.
Printed by Editor Bren, Bucharest, Romania (2007). (in Romanian)
29. Besnea, D., Rizescu, C.I., Rizescu, D., Comeaga, D., Ciobanu, R., Moraru, E.: Study of
deflection behavior of 3D printed leaf springs. In: IOP Conference Series: Materials Science
and Engineering, vol. 444, no. 4 (2018)
30. Cerda-Avila, S.N., Medellín-Castillo, H.I., De Lange, D.F.: Analysis and numerical
simulation of the structural performance of fused deposition modeling samples with variable
infill values. J. Eng. Mater. Technol. Trans. ASME 141(2)
31. He, L., Peng, H., Land, J., Fuge, M.D., Froehlich, J.E.: Designing 3D-printed deformation
behaviors using spring-based structures: an initial investigation. In: UIST 2017 Adjunct -
Adjunct Publication of the 30th Annual ACM Symposium on User Interface Software and
Technology, pp. 151–153 (2017)
32. D’Silva, S., Jain, S.: Design of a modified leaf spring with an integrated damping system for
added comfort and longer life. IJRET: Int. J. Res. Eng. Technol. 3, 30–34 (2014)
An Introductive Training for Welding
Workshop: A Biometric Evaluation Using
Virtual Reality Scenes to Improve Practice
Francisco Torres-Guerrero(&), Leticia Neira-Tovar,

and Luis Torres-Treviño
Universidad Autónoma de Nuevo León, Pedro de Alba S/N, Niños Héroes,

San Nicolás de los Garza, Nuevo Leon, Mexico
francico.torresgrr@uanl.edu.mx
Abstract. Welding is one of the most used process in manufacturing industry.

Training in this process demands a long period of time and involves several
resources such as specific facilities, metal pieces, energy, electrodes, gases, etc.
Evaluations requires those training conditions, which every day are needed in
welding industry. At the same time, this includes the safety understandings that
welding training implies. Considering these situations, this work proposes a
Virtual and Augmented Reality cabin, for evaluation and training in welding
process. The method used includes understandings of safety, knowledge of
workshop elements, welding tools identification and practical exercises to test
several welding skills. Testing was applied to an experimental group who
trained using the virtual welding tool, allowing the study of cognitive effects
caused by virtual training before any previous real practice; there was also was a
control group which trained only with real practices. By using the biometrical
device EPOC of Emotiv, concentration and stress in both groups while doing the
first welding real practice was tested. Experimental group showed less stress and
more concentration than control group.
Keywords: Mix reality Welding workshop Virtual training

Biometric evaluation
1 Introduction
The virtual and augmented reality program from Faculty of Mechanic and Electric
Engineering (UANL) has worked in the development of virtual tools for the training in
different industry processes, focusing in studying the design and construction of virtual
environments like workshops in order to optimize user experience, decrease risks such
as accidents, as well as resources costs. By using new technologies of the educational
software development, researchers aim to improve user experience. The use of new
technologies does not guarantee a better user performance, that for, it is necessary to
fulfill an exhaustive evaluation to find opportunity areas within the graphic interface
development. To measure accuracy of a virtual learning environment it is required to
make sure all elements are included [1].

https://doi.org/10.1007/978-3-030-18715-6_27
320 F. Torres-Guerrero et al.
There are research papers which indicates that using the information of technology
can improve the effectiveness of the learning in this field, using tools such as virtual
campuses, educational administrative platforms, as well as multimedia format tools
focusing on the improvements of academic content. In many occasions this is all
validated by performance of students [1, 2].
Welding is a key process in manufacturing industries. Nowadays, welding still
required when assembling metal structures. Those structures may be used in buildings,
or when assembling a car, a ship or other different kinds of vehicles. However, there is
plenty applications for welding in several industry areas. Some of these industries, such
as automotive industries, used robots in many welding processes they require. On the
other hand, in some other industries such as shipbuilding, heavy equipment production,
and small parts fabrication, welding is still a process done by human hands [3]. This
applied by human welding requires training, which is usually expensive. For example,
it was estimated that welding training cost for all U.S. shipyards was around $5M [4].
High costs in training caused industries which require welding look for alternative
training methods.
From the most recent projects with several publications, the industrial virtual
laboratory for welding has allowed the user to develop basic skills through human-
computer interaction. Welding needs of the user to develop different skills related to the
use of a highly dangerous industrial tool due to the use of combustion [5, 6].
By using VR technologies, is now possible to create real situations in many pos-
sible scenarios, and then test user skill for that situations. These technologies allow the
development of many low-cost solutions for different task or process such as welding.
VR solutions seems as a viable option for training in task which involves performing a
procedure [7]. One of the most important aspects of graphic interface development, is
to consider the success principles of usability within the design of interaction oriented
to user responsibility and environment [8].
This work aims to demonstrate emotional benefits for new employees during the
process of training, through the use of virtual tools as a support for training on man-
ufacturing process activities, using the case of welding training. This objective involves
not only virtual reality technologies, but also augmented reality. There is also a general
approach of design and manufacture an integral platform prototype that uses virtual and
augmented reality, internet of things and data mining, to improve training processes of
the welding machine operators within industry [9]. Training those skills is needed in
many important areas, from conduits in food and pharmaceutical industries, to many
applications in automotive industry. Achieving the creation of a smart device that will
transform training in companies and integrate them in 4.0 Industry Strategy.
2 Integrated Review
There are many commercial products like this project for operators of welding
machines that help them in the induction and certification processes, nevertheless, those
are very expensive. In certain cases, those technologies do not emulate realistic envi-
ronments: some neither detect arms movement and position nor posture of operators,
some are not easy to operate for the user or costumer, and some neither have biometric
An Introductive Training for Welding Workshop 321
sensor nor augmented reality module. On the other hand, there are technologies in
research phase, those which use methodologies based on sophisticated Computational
Vision Systems and Inertial Measurement Units sensors for the estimation of some
components of the orientation and position trajectory in some tools or applications,
some of those for welding [10].
Previous studies in welding skills suggest measuring those skills in terms of
position and speed. A skilled welder has a better control of position and speed of the
torch while an unskilled welder shows variations in both parameters. Those studies
propose an evaluation by using thresholds, in which a score depends the degree of
violation of the threshold [11].
Even though technologies based on Computational Vision Systems presents posi-
tive results in an internal and controlled environment, there are still narrow circum-
stances like shades, light interruptions, distance limitations and light interference,
which require of specialty software of images and videos, also, in certain cases,
expensive in economic terms [12]. Regarding the graphic environment, the current
virtual reality coaches present extremely real simulated environments, nevertheless, fail
the transition to real welding, measured and observed through augmented reality.
Biometric analysis usually consists of recording and electroencephalography
(EEG) signal, which is the function of the Emotiv EPOC headset. These system for
detecting and measuring biosignals consist of a set of sensors calibrated to detect those
biosignals from a user, as well as a system where to see results. Those biosignals can be
related to emotions, like concentrations and stress, which are studied in this work. In
product manufacturing, there are many extensive manual processes, which involves
heavy mental loads for workers. In order to avoid a faulty performance that may cause
accidents, it is necessary to monitor their mental state [13].
3 Method
The method consist on three steps, the first step was to develop a training tool based on
a videogame structure, which simulate a welding workshop that allows the users to
interact with different elements to get the basic knowledge and introduce them in
welding process.
The second step, to design an ocular test to identify attention.
Even in following instructions in traditional training, students may skip some
details, and from then on they may trigger accidents or damage workshop tools. The
advantages are reduced risk of accidents, avoid welding material high cost and to get a
preliminary evaluation of the user. While the time of activity in the VR workshop, the
student can interact in a safe way with welding material. Through interaction study it is
possible to know the time needed to complete training in a virtual course before to start
the real building practice.
The third step was the design of a biometric test, to know the stress level.
The final product consists in a training station whose hardware will be composed of
three forms of interaction:
(1) Training virtual reality mode (video game) by using a control which simulates a
welding torch, that user will control in tutorial and training mode while doing
virtual tasks.
(2) Evaluation through software of the welding tasks objectives.
(3) The eye tracker used to identify attention and ocular following.
(4) A Biometric sensor to know the stress and concentration level.
3.1 Materials
The tools used for platform design, application and test are indicated in the next list:
• 3ds max software. This software allowed the design of the different objects used in
the interface.
• Unity3D. Software used in virtual platform development. This software allows the
design of 3D scenes, such as the VR interface requires.
• Acer Mixed Reality Headset. This device is used to see the different virtual
scenarios of the platform. This headset is the one used when playing with the
interface.
• Acer Mixed Reality Controllers. Controller provided along with the headset.
These controllers are used to execute different events inside the VR platform, such
as pressing a button, or simulate a tool.
• Windows Mixed Reality Software. Software required to use both headset and
controllers. This software requires a Win10 compatible computer.
• SQLite software. This software is used to create a data base for storing information
of the interface. The data base contains all fields related to user data in registration
section, as well as performance data once exercises are done.
• Emotiv EPOC. This headset is able to collect brain data while user interacts with
the interface. This headset is used when testing the workshop so record data about
emotions.
• Eye Tracker. This device is composed of pupil movement sensors and is used to
capture date. Its function is to analyze the interest of people through what they see
and gives another approach when interpreting results.
• Gazepoint Control. Software provided by the Eye Tracker provider. This software
is used to calibrate sensor and cursor, as well as to select the screen where Eye
Tracker is working.
• Gazepoint Analysis. Software provided by the Eye Tracker provider. This software
reproduces and edit videos, images and screen recordings.
3.2 Interface Development

The proposed interface was designed using Unity 3D software. The whole interface is a
combination of 2D and 3D scenes, focused on the previously mentioned forms of
interaction. This interface consists of a menu shown in Figure, which is divide in four
different options, as shown in Fig. 1.
Fig. 1. Interface menu.
Informative Videos. This section works as a documentation for practical exercises.

This documentation is shown in the form of different videos, according to each exer-
cise. This section is shown in Fig. 2.
Fig. 2. Informative video section.
Quiz. In the quiz section, the user required to enter their data (name, id) and then
answer several questions, as shown in Fig. 3. By each well answered questions, the
user get 10 points and final scores are connected to the data base of the interface.
Fig. 3. Quiz section.

Practice. In here, user skills are tested. Once the user saw the informative videos
section, it is time to test what they learned. This section contains exercises which
involves welding cases exercises as well as material and working areas identification as
shown in Fig. 4. All the exercises scenes are connected to a database, so information
about time and performance is stored for later viewing.
Fig. 4. Practice section (Welding workshop).
Scores. In this section, data base collected information is shown. Once the users are
done with all the different exercises and the quiz, they can see their performance shown
as a grade. This section is shown in Fig. 5.
Fig. 5. Score section.

3.3 Evaluation Process

The evaluation process consists of 2 tests, first the eye-tracker test and then the bio-
metric test. When testing, is applied a thermal image analysis of eye tracking through of
an evaluation methodology for clustering. The eye tracker has been widely used to
evaluate attention and immersion issues since it allows to locate points of interest. For
the evaluation of tool used for welding training on virtual format, is looked for the
points in order to identify where attention of the user is. These points of interest work
as feedback because with that information is possible to see and analyze if there is any
distraction inside the interface.
To doing the evaluation is necessary to use the scenarios where the interaction
occurs as a basis and divide the main scenes by quadrants. In the next image the user
has the objective of identifying the tools for work out a welding. Here evaluation needs
to identify the attention areas in that moment to do the assigned task in order to
determine user experience.
The application architecture team identifies by means of cluster the areas related to
available tasks so define if the focusing point is a distractor point or it is related to the
task. Clusters where there are options to perform tasks are indicated in Fig. 6.
Fig. 6. Clusters area and eye tracker focus point.
The environment is prepared by implementing the eye tracker to identify the focus
points while performing the task assigned. In this example the task is to verify the
controls of the welding machine. In Fig. 7, it is possible to observe by a thermal
representation the focusing points of the eye tracker while performing the task.
Once the image is identified, it is divided into clusters where it is possible to locate
attention positions of user. This will allow evaluation with the use of the previously
structured template with clusters related to the task.
Fig. 7. Location of attention patterns
It was compared between the thermal marks of eye tracking with the templates
defined by the experts. This allows to check if the user was able to identify the cluster
with a task or it focused on a cluster that could result in being a distraction, as shown in
Fig. 8.
Fig. 8. Distracting areas.
On the other hand, there were other aspects to evaluate while using the interface.
These aspects are related to emotions, which are important because welding requires
user is focused on the task that must be done. The evaluation was done by using a
biometric headset. These headset records an EEG signal by the 16 electrodes which
composed the headset.
This experiment involves an experimental group, which trains with the virtual
welding tool before any real exercise, and a control group which trains only with real
exercises. This evaluation was applied to a total of 30 participants, 15 per group. All of
them realized a real welding practices for the first time, while using the EPOC headset.
In Fig. 9, there is a participant using the headset while doing a welding exercise.
Fig. 9. Training with VR platform and real welding practice.
The objective of this evaluations was to analyze through the biometric sensors in
the headset if there exist a better adaptation to the task when there is a previous VR
training. In this case, emotions concentration and stress were tested. During the
welding factor, a key factor in order to avoid accidents and perform all different task is
that user is well-focused on the task.
4 Results Analysis
To test the tool, the following learning activities shown in Table 1 were designed, from
which the user experience will be evaluated through eye tracking and a thermal analysis
will be made by means of clusters.
Table 1. Activity table.

Id Activities
1 /Safety equipment identification
2 /Butt preparation
3 /Base material preparation
4 /T Union preparation
5 /Welding machine calibration
In the process of analysis, 10 students, who did not have any previous experience in
welding participated. They received a detailed explanation of the activities to do. Each
participant carried out the activities for 3 different practices to observe the learning
process. Tests are summarized in Table 2.
Table 2. Testing observation.

User Practice 1 Practice 2 Practice 3
%Right %Distractor %Right %Distractor %Right %Distractor
1 50 50 70 30 90 10
2 65 35 80 20 100 0
3 70 30 90 10 100 0
4 30 70 80 20 96 4
5 60 40 90 10 80 20
6 40 60 60 40 95 5
7 50 50 80 20 90 10
8 67 33 85 15 100 0
9 70 30 80 20 90 10
10 75 25 85 15 100 0
In order to evaluate the effects of virtual training in emotions of the workers, a test
was designed with biometric sensors, using brain measuring technology, to identify if
there is a better adaptation in the environment when a previous training has been
carried out.
Two aspects of the EEG tool are evaluated, which are concentration and stress.
Within the welding process, a crucial factor in avoiding accidents and performing a
welding task is that the user is well focused on the task to be performed.
The same activities mentioned in Table 1 were applied for this test, to both
experimental and control group. Results are shown in Tables 3, 4, 5, 6 and 7.
Table 3. Safety equipment identification

Previous VR Non previous
trained trained
Stress Focus Stress Focus
Average 38 35 48 20
Min 15 21 30 17
Max 45 42 75 30
Std deviation 8.69 7.69 11.64 5.80
Table 4. Butt preparation

trained trained
Average 36 60 45 45
Min 27 26 39 38
Max 45 65 55 50
Std deviation 8.85 14.94 9.97 7
Table 5. Base material preparation

trained trained
Average 43 47 56 20
Min 33 40 23 15
Max 50 50 60 25
Std deviation 7.544 6.132 10.30 5.19
Table 6. T Union preparation

trained trained
Average 37 45 55 28
Min 33 37 50 22
Max 45 50 62 33
Std deviation 4.51 7.2 6.8 5.57
Table 7. Welding machine calibration

trained trained
Average 32 29 58 22
Min 20 17 40 17
Max 40 35 65 24
Std deviation 8.066 9.96 10.88 2.88
5 Conclusion
The present research allowed us to evaluate the welding tool at a usability level in a
welding station for the training of welding machine operators with the use of virtual
reality and augmented reality. Companies in Mexico that acquire it, at a reasonable
price, will achieve their training process while integrating to the strategy of the 4.0
Industry, which provides them a competitive advantage with respect to the companies
abroad. This will achieve a considerable economic spill for the country since it will
become the most important manufacturing country in the welding area with the use of
the virtual welding tool.
In addition to this, and continuing with social benefits subject, the fact that a
smaller amount of waste and welding machines are used will have an impact on the
environment by emitting less greenhouse gases and less waste.
There was a training experimental group, which trained using the virtual welding
tool previous to a real welding practice, and a training control group which trained only
with real practices. By using the Emotiv headset, concentration and stress where tested
in an initial real welding practice. Experimental group show less stress and more
concentration than control group. The use of virtual tools previous to a real practice
allow a cognitive preparation for the user, which is presented as economical, social and
safety related improvements. Those improvements will be teste don a future retailed
value study.
As for educational benefits, we will continue to work with the Autonomous
University of Nuevo Leon on advanced development relevant to the project, that focus
on the real needs of the industry and achieve to create innovative solutions for it. At the
same time, the device can be acquired by universities and technical high schools to
reduce the risks of accidents within them, achieving as well, trained soldering tech-
nicians ready for the labor market.
Acknowledgments. The research work reported here was made possible by participation of
Graduate and Under-graduate students at the Engineering School: Antonio Escobar, Oscar Leal,
Aldair Leon, Fredy Lucho and Edson Castañeda.
References
1. Torres, F., Neira, L., Del Rio, M.: A learning evaluation for an immersive virtual laboratory
for technical training applied into a welding workshop. Eurasia J. Math. Sci. Technol. Educ.
13, 521–532 (2017)
2. Kiritsis, D., Bufardi, A., Mavrikios, D., Knothe, T., Szigeti, H., Majumdar, A.: A
competence-based industrial learning approach for factories of the future. Educ. Inf.
Technol. 18(2), 331–350 (2013)
3. Porter, N., Allan, J., Gifford, T., Virtual reality welder training. Session 5: joining
technologies for Naval applications (2007)
4. Boutwell, R.: Welding training modernization. In: SNAME Ship Production Symposium,
Boston (2002)
5. Chen, S.J., Huang, N., Liu, Y.K., Zhang, Y.M.: Machine assisted manual torch operation:
system design, response modeling, and speed control. J. Intell. Manuf. 28(6), 1249–1258
(2017)
6. Mavrikios, D., Karabatsou, V., Fragos, D., Chryssolouris, G.: A prototype virtual reality-
based demonstrator for immersive and interactive simulation of welding processes. Int.
J. Comput. Integr. Manuf. 19(3), 294–300 (2006)
7. Grajewski, D., Buń, P., Górski, F.: Examination of effectiveness of a performed procedural
task using low-cost peripheral devices in VR. In: Virtual, Augmented and Mixed Reality:
Interaction, Navigation, Visualization, Embodiment, and Simulation 2018, pp. 403–415
(2018)
8. Zhang, M., Zhang, Z., Chang, Y., Aziz, E.-S., Esche, S., Chassapis, C.: Recent
developments in game-based virtual reality educational laboratories using the microsoft
kinect. Int. J. Emerg. Technol. Learn. 13, 138–159 (2018)
9. Borsci, S., Lawson, G., Jha, B., Burges, M., Salanitri, D.: Effectiveness of a multidevice 3D
virtual environment application to train car service maintenance procedures. Virtual Reality
20(1), 41–55 (2016)
10. Ni, D., Yew, A.W.W., Ong, S.K., Nee, A.Y.C.: Haptic and visual augmented reality
interface for programming welding robots. Adv. Manuf. 5(3), 191–198 (2017)
11. Erden, M.S., Tomiyama, T.: Identifying welding skills for training and assistance with robot.
Sci. Technol. Weld. Joining 14(6), 523–532 (2009)
12. Xie, B., Zhou, Q.: Yu, L.: A real-time welding training system base on virtual reality. In:
2015 IEEE Virtual Reality (VR), pp. 309–310 (2015)
13. Xiao, H., Duan, Y., Zhang, Z., Li, M.: Detection and estimation of mental fatigue in manual
assembly process of complex products. Assembly Autom. 38(2), 239–247 (2018)
Simulation of Flexible Manufacturing Systems
as an Element of Education Towards
Industry 4.0
Rafał Kluz(&) and Katarzyna Antosz
Faculty of Mechanical Engineering and Aeronautics,

Rzeszow University of Technology, Rzeszów, Poland
rkktmiop@prz.edu.pl
Abstract. Industry 4.0, known as the fourth digital revolution, is the integration
of modern systems and machines based on digital solutions that influences the
productivity of an enterprise. This concept forces the use of more and more
sophisticated and complex solutions by enterprises, i.e. technological or orga-
nizational, associated with data collection and analysis, the development of
technology, the ability of a rapid response to both internal and external changes,
and the improvement of present processes. Highly automated production pro-
cesses will require qualified management staff as well as production workers
experienced in the work with new materials, machines and, in particular,
information. This work presents the course of a didactic process which aims at
the acquisition of skills within the simulation and analysis of flexible manu-
facturing systems with the use of mass service systems theory. The main part of
the work shows the course of classes which aim to study the work efficiency of a
flexible manufacturing cell. As a criterion of the system work efficiency, basic
indicators of the mass service system work were adopted, that is an average
number of products waiting in a queue, average time of a product waiting in a
queue and a manufacturing system as well as the expected downtime of a cell.
The results were compared with the results of typical models describing the
work of mass service systems (M/M/1, M/D/1, M/El/1, M/G/1).
Keywords: Industry 4.0 Flexible manufacturing system

Simulations of manufacturing systems work Mass service system
Didactic process
1 Introduction
In principle the concept of Industry 4.0 is supposed to forward the transformation of

manufacturing enterprises into smart factories in which the networks based on infor-
mation and communication technologies join machines, processes, systems, products,
clients and suppliers [1, 2]. A smart factory is to be the final effect of this most complex
of industrial revolutions. In such a factory, intelligent networks join machines, pro-
cesses, systems, products, clients and suppliers. Due to this fact further advancement of
automation, continuous optimization of products and processes, collection and pro-
cessing of a huge number of data in the real time, preventive machine maintenance and

https://doi.org/10.1007/978-3-030-18715-6_28
Simulation of Flexible Manufacturing Systems 333
devices as well as fast adaptation to the changes on the market will be possible. The
fourth industrial revolution is possible due to the new technologies development [3].
The new concept of industry concerns such areas of an organization operation that are
aided by intelligent systems facilitating the decision making and automation that
improves productivity and work quality. It requires significant mechatronization of
products, devices, machines and the systems of their production [4]. In connection to
this, the attempts of creating new forms of production organization are undertaken
according to the evolutionary advance within automated manufacturing systems, up to
the self-regulating automated manufacturing cells (SAMC) and self-transforming
automated manufacturing systems (SAMS), what was presented in the works [5].
Flexible Manufacturing System (FMS) is a computer integrated complex of
numerically controlled machines and technological devices (machine tools, robots)
which allow to use different manufacturing techniques (machining, heat treatment,
forming process, washing, drying, etc.) with a wide use of measuring and diagnostic
devices joined together with automated means of transport (transporters, self-propelled
fork-lift trucks) ensuring automated manufacturing of goods of diverse construction
features [5–7].
Flexible Manufacturing Systems modelling is a difficult and time-consuming task.
In order to conduct the analyses of flexible manufacturing systems work, simulation
and analytical models may be used. Simulation models are constructed on the basis of
special programming languages and systems. They allow to analyse complex systems
despite the fact that they don’t possess a sufficiently large basis of the results obtained.
However, they require computers with high computing power [8, 9].
On the grounds of the complexity of the problem of the flexible manufacturing
systems work simulation and the need of its analysis that derive from the Industry 4.0
concepts, the course of Flexible Manufacturing Systems has been introduced into the
university syllabus at Rzeszow University of Technology.
2 Scope and Aim of Education
The industry 4.0 concept requires the company to respond quickly to customer
requirements. This means that the manufacturing system should enable the production
of priority products that may reduce system performance and cause disruptions to its
operation. To make this possible, the company should have some reserves of pro-
duction capacity. Too large reserves, however, reduce the efficiency of the production
system and therefore should be set very precisely. In an enterprise based on the concept
of industry 4.0, you can use digital data, often stored in the cloud, to determine the
actual production capacity of the processing sockets and, accordingly, redirection of
priority declarations (products).
However, the collected data must be finally analyzed, and on its basis a physical
model should be created that gives the possibility to control the process. In cyber
physical systems, the collected data is compared to the model and feedback is gener-
ated allowing modification of the material flow process in order to achieve the desired
results. The process is not controlled solely by the operator, but also a computer that
analyzes and compares the data relative to the model in real time. Therefore, the model
334 R. Kluz and K. Antosz
must accurately describe the elements of the process. Creating such a model requires a
full understanding of the system’s work.
The course Flexible Manufacturing System (FMS) is realized for the students of the
second year in the field of Management and Production Engineering with specialization
in Integrated Manufacturing Systems. The main aim of education within this course is
to provide students with the basic theoretical knowledge in the construction and
operation of flexible manufacturing systems as well as the acquisition by a student of
the skills of designing a spatial structure of a flexible manufacturing system and the
control of a production flow in flexible machining and manufacturing cells. A student
who takes up classes of the course should meet the requirements determined in the
category of knowledge/skills/social competences. Preliminary requirements in the
Knowledge category oblige a student to be familiar with the knowledge basics of the
automation of manufacturing processes and machine technology. Preliminary
requirements in the Skills category require a student to possess team work skills.
Preliminary requirements in the Social Competences category state that a student need
be able to broaden their knowledge and improve their professional skills individually.
The course is realized in lectures (15 h) and laboratory classes (30 h). The topics of
the lectures include: introduction to flexible manufacturing systems (FMS), forms of
manufacturing organization of FMS, characteristics of subsystems in FMS: manufac-
turing subsystem, machining products flow subsystem, storage subsystem, manipula-
tion subsystem, production control in FMS, models and methods of design and
production control in FMS, mass service network models, Petri net models, simulation
models, and mathematical programming models. Laboratory classes include the fol-
lowing topics: developing a simplified technological process for a group of techno-
logically similar parts, prioritizing of tasks in a flexible machining cell, the
configuration of a flexible robotic machining cell, the selection of a kinematic structure
of an FMC manipulation subsystem, the selection of a material flow subsystem and the
analysis of the effectiveness of a flexible machining cell work with the help of mass
service systems.
The aim of the classes conducted by students is to identify the model of mass
service system. Then, for a given level of intensity of applications, based on the
developed computer program, students have the opportunity to determine the reserve
capacity of the nest ensuring reliable operation. The results of the calculations are
verified in the actual production conditions in the laboratory, while the test results are
used to enrich the computer program database. The learning outcomes were determined
as follows:
1. The student has deeper knowledge concerning the construction and operation of
flexible manufacturing systems (FMS), and the methods and tools that allow to
describe FMS structures and the processes within them.
2. The student is able to assess a technical and organizational solution of a flexible
manufacturing system in terms of effectiveness and operation reliability.
3. The student is able, according to a given specification, to design a configuration of a
flexible machining cell.
4. The student is able to design a manipulation and material flow subsystem with the
use of appropriate methods, techniques and tools, using the existent tools or
inventing new ones.
5. The student is able to conduct the research concerning the work effectiveness of the
integrated manufacturing systems.
6. Determining the capacity reserves of the sockets ensuring reliable operation.
7. The results verification in the real machining socket.
3 Analysis of the Effectiveness of the Flexible Machining Cell

Work with Mass Service Systems
3.1 Scope and Aim of Laboratory Classes

One of the topics realized in laboratory classes is the assessment of the effectiveness of
flexible machining cell work with the use of mass service systems. Classes are realized
in the following stages:
1. Theoretical basis – explanation of mass service systems.
2. Determining the configuration of a flexible machining cell (FMC) (the choice of a
machine tool, manipulation subsystem, transportation subsystem).
3. Assigning product components into groups (families) according to their features
and design and technology similarities in compliance with the rule of “group
technology”.
4. Establishing technological processes of the parts for processing in FMC with the
use of a chosen CAM system.
5. Determining basic parameters that describe the work of service channels (of a cell).
6. Determining the indicators of the effectiveness of the cell work with the use of
typical models describing mass service systems work (M/M/1, M/D/1, M/El/1,
M/G/1).
7. Simulating the cell work (measuring basic parameters: an average number of
queueing products, average time of a product in a queue and in a system, expected
downtime of a cell).
8. Assessing the relevance of a mathematical model.
9. Analyzing the cell work with the use of a chosen mathematical model.
10. Determining the minimal production reserve capacity, assuming a possible
appearance of a priority inflow of the arrivals on the intensity ki.
The aim of these classes is to gain skills of the assessment of the flexible manu-
facturing system work through determining which of the presented models of Mass
Service Systems is responsible to the greatest extent for the work of the flexible
machining cell models designed by students. Classes are divided into two parts, as
follows: theoretical introduction and practical classes.
3.2 Theoretical Basics – Mass Service Systems

The practical part provides a student with models based on the mass service theory that
belong to one of the most effective analytical models describing the work of flexible
manufacturing systems [10]. Mass service theory (queueing theory) is an autonomous
field of knowledge based on mathematical statistics and probability theory. It analyses
the systems in which queues arise. The method allows to assess the effectiveness of the
systems work by means of such parameters as: probability that a new arrival will need
to wait for service, average waiting time for service or an average queue length, etc.
such information allows designers of flexible manufacturing systems to compare dif-
ferent system variants as well as to assess the profitability of investing into additional
service stands [11, 12].
The abbreviated labelling of mass service systems, sequences of arrivals and ser-
vice sequences uses the notations introduced by D. Kendall, in which the system is
characterized by the ordered five parameters X/Y/m/L/k, where [10, 11]:
– X is the distribution of a random variable of an arrival,
– Y is the distribution of a random variable of task service time,
– M is a number of service stands,
– L is the capacity of the length of que (waiting space),
– K is the dimension of a calling source.
If L = k = ∞, thus Kendall’s notation is limited to the three first fields. The dis-
tribution of random variables is described by the following symbols [5]:
– D – deterministic or fixed (single – point distribution) stream,
– M – exponential distribution of service times or intervals between arrivals, or
Poisson distribution,
– Ek – The Erlang distribution with parameter k, which may appear both in service
machines as well as in arrivals,
– G – the stream of discretionary time distribution.
According to the presented notation: M/M/1 is a mass service system with a single
service channel for which the arrivals entry stream is described by Poisson distribution,
and the service time by the exponential distribution. On the other hand, a system
described by the notation of M/D/1 is characterized by deterministic (one-point) arri-
vals service time [6].
The systems presented determine border ranges which include the values of basic
parameters of the systems with discretionary different distribution of service time but
identical intensities of arrivals k inflow and service l. This case is very clearly illus-
trated by the graph presenting the queue size (an average queue length) with different
distributions of service time depending on the value of the system load parameter
(Erlang constant q).
The aim of such an analysis is to determine an optimal number of service stands or
the rule governing the choice of arrivals waiting in a queue for service. Providing the
established average waiting time for service or a certain queue length may be the result
of the analysis. Due to the fact that exact times of a task arrival to the system, as well as
task service times by the system are not known a priori, it is assumed that they are
random variables. Thus, they are analysed with the help of nondeterministic models.
However, these models require to define a few basic parameters such as: the character
of the probability distribution describing the stream of arrivals and service as well as
the queue discipline. The results obtained on their basis are characterized be a relatively
high precision, although dependent on the precision of estimation of the basic system
parameters. Nevertheless, as it is known, the real parameters of the distribution differ
considerably from the models. This fact affects the precision of predicting the systems
work effectiveness [literature] (Fig. 1). In practice, in order to describe the work of a
flexible machining cell, a few typical models that describe mass service systems work
may be used (M/M/1, M/D/1, M/El/1, M/G/1). Models with the exponential (M/M/1)
and deterministic (M/D/1) service time determine the ranges that include the values of
basic parameters of the systems with discretionary different distribution, but with
identical intensities k and l.
Fig. 1. An average queue length in the chosen systems.
It may be exemplified by the graph of a queue size, at different service time

distributions, depending on a load parameter q, Fig. 1.
3.3 Mass Service Systems – Classes

The aim of the classes is to establish which of the presented models of Mass Service
Systems corresponds to the machining cell designed by students. In order to realize the
classes, the following assumptions were made: only the parts technologically similar
shall be made in a system (assigned to a cell according to the rules of “group tech-
nology”) and characterized by different calculated times, in any moment an order to
manufacture any product within the system possibilities may be made, whereby the
system won’t need a changeover. As a result, it will be characterized by high flexibility
of the kinds of products and the following parameters: the stream of arrivals, queue
discipline and arrivals service.
The stream of product arrivals that is the stream in which the time interval (t) be-
tween next arrivals is a random variable of an exponential distribution (or a number of
arrivals in a time unit has a Poisson distribution) of a probability density function (1).
f ðtÞ ¼ kekt dla t 0 ð1Þ
with an expected value (2):
1 1
E ðt Þ ¼ ¼ ð2Þ
k 6:1
In order to conduct the experiment for the inflow of arrivals, a sequence of pseu-
dorandom numbers of a given exponential distribution is generated. Figure 2 shows a
histogram of the distribution of a random variable for the arrivals inflow.
40
Number of applications after the time t
35
30
25
20
15
10
0
1,000 6,875 12,750 18,625 24,500 30,375 36,250 42,125 48,000
t[min]
Fig. 2. A histogram of a random variable for the arrivals inflow.
Additionally, it is assumed that “a natural queue” (FIFO – First In First Out), that
means the first flowing arrival in a queue, will be serviced first.
During the classes the assessment of the system work is done. There are 12, 18, 24,
36 technologically similar parts produced in this system. For the adopted groups of the
realized products, students compare the distributions obtained on the basis of the real
calculated times with the model distributions. Figure 3 shows the distributions of a
random variable of the service time for the adopted groups of the realized products.
A student should notice during the task that these distributions differ from the model
distributions, what affects the precision of predicting the chosen models.
During the classes the basic system parameters are also monitored. They include
the waiting time of each product in a queue, the time a product stays in a system and a
number of products (basic units) waiting in a queue. The measurement of a queue
length is done in constant time intervals t longer than the average service time l (t > l).
The analysis of the graphs, done by students, showing the results obtained on the basis
of the mathematical modelling and experiments, proves that the assessment of the
model consistency should be conducted considering the number of parts produced in a
system. During the further detailed analyses, students conduct an analysis of the service
time for the established product levels. They also assess the results. They notice that in
case with 12 different products manufactured in the system, the probability distribution
of service time vary from the typical known distributions. It is immediately reflected in
the precision of the results. The average queue length determined by experiment was
0.83 pieces, Fig. 4.
a) b)
35 26
24
30 22
Number of observations
20
25 18
16
20
14
12
15
10
10 8
6
5 4
2
0 0
3 4 5 6 7 8 9 10 11 3 4 5 6 7 8 9 10 11
t [min] t [min]
c)
26
d)
30
24
28
22 26
20
24
18 22
20
16
18
14 16
12 14
10 12
10
8 8
6 6
4 4
2 2
0
0 3 4 5 6 7 8 9 10 11
3 4 5 6 7 8 9 10 11
t [min]
t [min]
Fig. 3. Histograms of a random variable of the service time for a stand of: (a) 12 teams, (b) 18
teams, (c) 24 teams, (d) 30 teams.
Fig. 4. An average queue length in a system.
The result of M/E2/1 model was the closest to this value and equaled 0.79 with
4.82% error, on the other hand, the result based on the most commonly used model
M/M/1 had 27.7% error. Similarly, this is the case with the remaining indicators.
The average time of a product spent in a system determined by experiment was
14.4 min, whereas in a queue it was 7.2 min (Figs. 5 and 6). For that matter, in this
case the highest convergence with the study results was observed in M/E2/1 model
with the error respectively of 2.9% and 8.33% (model M/M1: −15.06% and 44.44%). It
means that a system designer, assuming an exponential distribution of service time,
obtains overstate values of the indicators of the cell work effectiveness, what may result
in incorrect decisions concerning the number of parallel service channels.
Fig. 5. An average waiting time of a product

Fig. 6. An average time of a product in a
in a queue.
system.
Assuming that there may be 18 parts manufactured in a system, the situation is

similar. In this case, students notice that the highest convergence with the study results
was observed with M/E2/1 model. The error values for an average length queue,
average time of products in a system and a queue were respectively: 7.69%, 9.51% and
12%. This error was characterized by higher values (−24.7%, −6.8% and 17.25%),
however, much lower than in case with 12 parts. The situation changed in case when 24
and 30 parts were assembled. Then, the highest convergence with the study results was
observed with M/M/1 model with the errors of: −9.23%, 5.30%, −16.5% (24 teams)
and −3.7%, 3.55% and 14.34% (30 teams).
The classes indicate that if a small number of parts is manufactured in a machining
cell, the adoption of M/M/1 model may not be justified in many cases. The higher
convergence with the results of the study realized during laboratory classes was
observed in this case with M/E2/1 model (a model with two Erlang phases). In case
when the number of parts in a system is increased (in the considered case more than
24), it is justified to use M/M/1 model. Orders for manufacturing the products with
longer calculated times appear then. As a result, service time is more comparable to an
exponential distribution but, at the same time, it worsens the system work effectiveness
indicators.
The analysis of the results is the last stage of the classes. After finishing them, a
student prepares a report.
4 Conclusions
The models based on the mass service systems theory are one of the most promising
analytical models that describe the work of flexible manufacturing systems. The
information gained on their basis allows managers to compare various organizational
types of a system, or assess the profitability of investment in additional service stands.
Their main disadvantage is their sensitivity to assess the basic system parameters
precisely. During the classes a student acquires the skills of using a simulation in order
to analyse the work of flexible manufacturing systems, which is one of the key areas of
Industry 4.0. As a result, a student is able to design a spatial structure of a flexible
manufacturing system, to categorize the components of products into groups (families)
according to their technological similarity, and to assign them to particular cells. On the
basis of the simulation study, a student is able to select an appropriate queueing model
that describes the work of the configured cell. As a result, it provides the opportunity to
determine an optimal number of service channels according to the criteria of the lowest
cost of the whole system operation (the cost of downtimes, the cost of an arrival loss).
Additionally students acquire the ability to respond quickly to changing market
demand, data analysis and simulation ability to acquire and process control of material
flow will let to choose the right way that ensures the reliability of the production system
in Industry 4.0.
References
1. Hermann, M., Pentek, T., Otto, B.: Design principles for Industrie 4.0 scenarios: a literature
review. Working Paper No. 01/2015. http://www.thiagobranquinho.com/wp-content/
uploads/2016/11/Design-Principles-for-Industrie-4_0-Scenarios.pdf. Accessed 29 Oct 2018
Machado, J.: Collaborative manufacturing based on cloud, and on other I4.0 oriented
Eng. Rev. 9(3), 90–99 (2018)
3. Stadnicka, D., Zielecki, W., Sęp, J.: Conception of Industry 4.0 – assessment of
implementation possibilities on the example the chosen company. http://www.ptzp.org.pl/
files/konferencje/kzz/artyk_pdf_2017/T1/t1_472.pdf. Accessed 28 Oct 2018. (in Polish)
4. Olszewski, M.: Mechatronization of product and production – Industry 4.0. Measur. Robot.
Autom. 3(20), 13–28 (2016)
5. Gola, A., Świć, A.: Computer-aided machine tool selection for focused flexibility
manufacturing systems using economical criteria. Actual Prob. Econ. 124, 383–389 (2011)
6. Świć, A., Taranienko, W.: Technological Design of Flexible Production Systems.
Wydawnictwo Politechniki Lubelskiej, Lublin (2003). (in Polish)
7. Afentakis, P.: A model for layout design in FMS. In: Kusiak, A. (ed.), Flexible
Manufacturing Systems: Methods and Studies, North Holland, Amsterdam, pp. 127–139
(1986)
8. Tolio, T. (ed.): Design of Flexible Production Systems. Springer, Berlin, Heidelberg (2009)
9. Kluz, R., Ciecińska, B.: Possibilities of simulation of production processes in a modular
mechatronic system. TiAM 3, 62–66 (2012). (in Polish)
10. Burduk, A., Chlebus, E.: Methods of risk evaluation in manufacturing systems. Arch. Civil
Mech. Eng. 9(3), 17–30 (2009)
11. Ayyappan, G., Sekar, G., Muthu Ganapathi Subramanian, A.: M/M/1 retrial queuing system
with vacation interruptions under ERLANG-K service. Int. J. Comput. Appl. 2(1), 52–57
(2010)
12. Loska, A.: Remarks about modelling of maintenance processes with the use of scenario
techniques. Maint. Reliab. 14(2), 92–98 (2012)
VR and AR in Lean Manufacturing Classes
Paweł Buń(&), Justyna Trojanowska, and Paulina Rewers
Faculty of Mechanical Engineering and Management,

pawel.bun@put.poznan.pl
Abstract. The paper presents preliminary results of research on integrating

Virtual and Augmented Reality in Production Management and Lean Manu-
facturing classes. The aim of the classes is to familiarize students Production
Management and Lean Manufacturing tools which are commonly used in many
manufacturing enterprises. Two applications (VR training application and AR
assembly instructions) were tested by a number of students of Production
Engineering. The results suggest that both applications can improve the process
of assembly operations as well as LM classes.
Keywords: Virtual reality Augmented reality Object manipulation

Lean Management Assembly
1 Introduction
One of the most prevalent professional application of virtual reality (VR) and aug-
mented reality (AR) is education and trainings – in engineering [1], school education
[2] or medical training [3]. In industry, VR is quite often used for support of design
processes, simulation of machines and devices or for decision-making [4]. VR and AR
are also used for industrial training purposes, some authors point out their importance
in the Industry 4.0 concept and factories of the future [5, 6]. VR and AR are especially
well suited to assembly process training [7]. The VR training allows increasing
effectiveness of knowledge transfer. It is far easier to repeat activities in VR, as well as
train undesired or dangerous situations. AR, on the other hand, allows training directly
on the production line, without prior knowledge of the process and helps in commu-
nication between a human and a machine.
Basic difference between AR and VR, which is visibility of a real world (present in
AR, cut out of VR simulations) makes for different, yet still complementary applica-
tions, especially in industrial training. With the emergence of low-cost VR and AR
interaction devices related with the interest of corporations related to the e-
entertainment, communication and image industries, VR technology became widely
available. Applications that use potential of latest interaction devices to the fullest will
allow users feel physically set among the virtual world or feel a way of presence, that
sets VR apart and takes traditional computing interfaces to the subsequent level [8].
Interactions and comparisons between use of AR and VR in manufacturing have not

https://doi.org/10.1007/978-3-030-18715-6_29
VR and AR in Lean Manufacturing Classes 343
been extensively studied so far, that is why it is justified to undertake industrial studies
on efficiency of use of both technologies, especially in training procedures.
The paper describes results of introducing the AR and VR applications in Lean
Manufacturing classes in Poznan University of Technology. AR application was used
as an interactive assembly instruction during 5S classes and VR training was intro-
duced as part of Lean Manufacturing workshops. In both cases applications were used
to improve assembly processes.
2.1 Problem and Aim Definition

Production companies use various strategies while competing for a strong position in
the market. Aside from improvement in manufacturing processes themselves – such as
Lean Manufacturing (see next chapter) – more general strategies can be employed. One
of them is readiness to fulfill individual needs of clients. This strategy is known as the
mass customization (MC) [9]. As opposed to mass production, which means manu-
facturing repeatable (and usually identical) products in high amounts, the mass cus-
tomization allows to fulfill expectations of each client as regards a given product,
known as a configurable or variant product. It does so by adjusting the product to his
individual needs. Its implementation is very attractive from the recipient’s viewpoint,
but it is very difficult for a production company and has a certain risk of failure,
especially as a result of increasing costs of design and manufacturing [10].
Mass customization is a problem from a viewpoint of training the employees,
especially in manual assembly processes. As each product can be different, it is difficult
to initially train employees and make them gain muscle memory. The virtual tech-
nologies (VR and AR) can help to a great degree – VR can make employees better
prepared for the task of being flexible before the production is commenced, while AR
can take away the need of preparing instructions at a given workplace – every infor-
mation on new product variant can be quickly sent to personal device of a given worker
to allow him to quickly react to changes in production plan.
The problem is that both technologies are relatively rarely studied together in the
same context of flexible production according to the MC concept. As such, there are no
elaborations on which strategy of implementation of AR and VR is justified in terms of
manual assembly of customized products. The aim of the preliminary studies presented
in this paper is to lay grounds for such a strategy, by introducing AR and VR in Lean
Manufacturing classes.
2.2 Lean Manufacturing

Organization of production flow conduct a crucial role in a company, especially in job-
shop production system, in which similar workstations are grouped together and flow
of processing materials are discontinuous because of needs to transport them from one
workstation to another [11]. Lean Manufacturing (LM) began in the 1950s in Japan,
more specifically in Toyota [12]. In order to eliminate waste in production techniques
344 P. Buń et al.
such as small batch production, set up reduction, inventory reduction, high quality
focus, and others were used. This new approach came to be known as the Toyota
Production System [13]. According to Liker [14] it allows to reach smaller cycle times
through the elimination of waste. In this case, the goal is to increase the time value that
is added during the production process, reducing the time between the customer order
and the final delivery. LM aims to reduce the time of process throughput, that is, the
time from when the raw material enters until it is incorporated in the final product.
Lean Management is an organizational production model that focuses on fostering a
philosophy of continuous improvement, eliminating wastefulness in the process,
reducing unnecessary costs, and thereby increasing productivity and customer satis-
faction. In other words, it is a system with fewer inputs to achieve the same objectives
as traditional mass production systems, but it offers a greater variety of products [15].
Lean manufacturing contains such tools as cellular manufacturing (CM), line bal-
ancing, value stream mapping (VSM), Inventory control, U-line system, single minute
exchange of dies (SMED), production levelling, pull system, kanban, standard work,
kaizen, one piece flow, poke-yoke, visual control etc.
One of the most widely used tools in LM is 5S, which name of the method comes
from the names of the five stages of organization of the workplace [12]. Sort, Set in
order, Shine/Sweeping, Standardize, Sustain.
2.3 Research Outline

The authors in previous studies have repeatedly undertook the problem of manipulation
of virtual objects [16–19] and interaction with a virtual environment to ensure the
highest possible level of immersion and, consequently, effective transfer of knowledge
[20].
This time, two different approaches were taken to introduce the AR and VR
technology in the Lean Manufacturing classes:
1. Virtual Reality application for training of a physical assembly process of a model
product out of modular blocks.
2. Augmented Reality application for introducing the 5S practice and help in the
assembly process of a model product out of LEGO blocks.
Two applications were built using the Unity engine and they were tested by a number
of students of Production Engineering. The tests involved a control group, which did not
use any of virtual technologies at all. No group used both approaches, to exclude
influence of use of one technology on the other one. Times of operation and numbers of
errors were recorded for each student, to evaluate effectiveness of each approach.
3 Applications Used in Classes
3.1 VR Training Application in LM Workshops

The main goal of Lean Manufacturing workshop classes is to show students that using
LM tools organization and production process can be improved by. Workshops classes
are carried out in the form of simulations in which each student plays a specific role,
and each role involves the need to perform specific tasks. Students perform assembly
operations on five workstations and provide to the “customer” a set of consistent
quality finished products at a given time. In order to improve the organization of the
process students observe the flow of materials and information between assembly
stations, measure basic parameters of assembly processes and analyze results.
During traditional classes, students receive paper instructions containing drawing
and information about how to assemble blocks and have 20 min to learn it. Unqualified
operator can artificially disrupt flow of materials and information. VR training appli-
cation can be used to prepare operators to work at assembly stations time, which in
result will increase the available time for process analysis.
The application was widely described in [21]. During the preliminary work the
application was prepared in two versions, differing in the amount of information
provided to user and the manner of presenting it. In both cases, in order to learn how to
perform the real assembly operation, the user had to properly manipulate virtual objects
as to submit a complete product. The additional task was to do it in the shortest possible
time.
The prepared solution uses low-cost VR devices with dedicated controllers (Oculus
Rift CV1 and HTC Vive) to manipulate objects in a virtual environment. In order to
Fig. 1. Assembly instructions in second version of training app [21].

346 P. Buń et al.
grab a block, the user had to put a virtual representation of his hand inside block (in
order to start internal event “collision”) and press “grab” button on controller, each time
the virtual hand was in the correct position relative to the block, the controller vibrated.
By moving and rotating the controller in reality user was able to manipulate virtual
hand and grabbed object.
Hints in first version of application were semi-opaque blocks that indicated where
user should put grabbed object and floating numbers indicating number of pins on each
sides. In second version the hints were removed, but user could look at assembly
instructions placed around the virtual work station (one of instructions presented in
Fig. 1).
Before starting VR training students had a few minutes to become familiar with
controllers and methods of interactions with virtual environment. The training itself
took 5 min, after that Lean Manufacturing workshops continued the in a traditional
way.
3.2 AR Application in 5S Classes

Application was widely described in [22]. It was introduced to classes in order to
familiarize students with the rules and stages of the 5S tool and possibility of using
Augmented Reality into assembly process. Students were divided into groups, 2–3
people in each. Classes were divided into 3 passes and, in each subsequent pass the
workstation was better organized thanks to using 5S method. In each phase students task
was to assemble simple product with 14 parts, each group had to submit a different
variant of the product, differing in the colors of individual blocks. After each students
changed workspaces (and product at the same time). In the first pass students had only a
picture of the finished product (as shown in Fig. 2) and a mix of different Lego blocks.
Fig. 2. View in app before entering code responsible for product variant [22]
First task was to assemble product in shortest time possible, after which, each group
perform the first 4 stages of 5S. The last task during first pass to create paper
instructions for assembling the product.
After changing workspaces each group had to assemble product as fast as they can,
using paper instructions made by previous team.
After changing workspaces once again students were given the sheet of paper with
AR marker. After scanning the marker students had to enter the appropriate code which
determined the product variant (Fig. 3).
Fig. 3. View in app before entering code responsible for product variant [22]
4 Results
4.1 VR Application
After VR training the students had to assemble product 10 times. In order to achieve
that they had to perform 5 operations for assembly. For each student the total duration
of ten operations of a given type and the number of errors was measured. It should be
noted that assembly operations requires different amount of focus. In operations 2, 3
and 5 only one block is attached and positioning of blocks doesn’t require counting of
pins. Operations no. 1 and 4 are far more complicated.
Students from first group (VR1) were trained in app with visual hints described
above. Second group (VR2) used training app with only assembly instructions. VR0 is
the control group, which undertook traditional training. In each group there were 20
students.
The analysis of Table 1 shows that students from the group 1 carried out their tasks
in shorter time in 4 of 5 cases, independently on the complexity of the operation. This
allows to state, that it is highly probable that training is more effective when a student
can familiarize himself with a virtual instruction beforehand, especially for more
complex processes. The average time of operations for both groups trained in VR
indicates that this type of training is effective for all tasks with this level of complexity.
Students from the first group (VR1) made 51% fewer errors than the control group,
as opposed to 40% in the second group (VR2) (Table 2).
348 P. Buń et al.
Table 1. Average time of assembly operations for each group after 10 repetitions
Parameter Group Average time of operations [s]
1 2 3 4 5
Mean VR0 217.3 197.7 167.8 168.4 139.8
VR1 149.3 133.3 67.2 74.3 72.2
VR2 171.5 116.5 81.7 102.8 103.0
Median VR0 244.5 172.5 134.5 157.5 126
VR1 154 134 60 81 72
VR2 180 101 65 94.5 86.5
Dispersion VR0 257 256 441 444 231
VR1 180 261 87 114 75
VR2 234 192 198 202 149
Table 2. Average number of errors for each group trained in VR

Parameter Group Number of assembly
errors
1 2 3 4 5
Mean VR0 3.9 3.6 0.1 1.3 2.6
VR1 5.3 1.9 0.0 1.3 0.2
VR2 3.1 3.2 0.4 1.0 2.3
Median VR0 4.5 2.5 0 0 1
VR1 6 2 0 0 0
VR2 2 2.5 0 0 0
Dispersion VR0 9 10 1 9 9
VR1 9 7 0 6 1
VR2 10 8 0 0 0
4.2 AR Application
The AR application was evaluated based on time of the task and questionnaire in which
students could give their opinions. The classes were taken by approximately 80 stu-
dents. The students were divided into groups, 2–3 persons in each group. Their average
results are presented in Table 3.
Table 3. Average time for each pass [22].

Parameter pass 1 pass 2 pass 3
Mean (s) 213 41 55
Median (s) 187 28 32
Dispersion (s) 364 106 200
The passage 1 of the assembly took the longest time. As expected, usage of the 5S
tools helped reduce the time. The dispersion of results is very high, this is probably due
to problems with effective use of the AR application itself, probably because of its
unintuitive interface. Results in Table 4 confirm this.
Table 4. Survey given to students after testing application. Mean values in second column [22].
Question Scale
1–5
How do you evaluate the overall applicability of the application (in the context of 3.35
the discussed issue)
How do you evaluate the interface’s intuitiveness 3.5
To what extent the application of augmented reality facilitates the correct 3.625
execution of the task
How important is the ability to rotate an object using the rotate function 4.125
How intuitive was the assignment of functions to the buttons on the screen 2.95
5 Summary
The presented research prove that both AR and VR applications can be used to improve
quality of Lean Manufacturing classes. It should be noted, however, that there is a lot to
improve especially in terms of ease of use of both applications.
The results show clearly, that VR helps in faster performance of assembly opera-
tions and the more complex process, the more it helps. The AR was also proven to be
effective in reducing errors and shortening time of performing the assembly operation.
Both technologies could be implemented together to achieve a synergy effect, which
will be studied in the future by the authors. In both technologies, flexibility of an
assembled product can be implemented freely, so each trainee can have his own version
of a given product.
In future development and studies, a real production line will be introduced into the
studies, to test interactions between VR and AR, as well as learning transfer differences
between VR/AR and control groups. The GUI and interaction will be improved after
surveys conducted in the initial studies, presented in this paper.
The virtual technologies will surely become more common both in industry and
daily life along with the increase in computing power of computers and mobile devices.
Availability of hardware and its capabilities in natural interaction and image projection
are higher and higher each year. Therefore, training processes in manufacturing
industries should be adapted to this. Possibility of constant employment of both AR
and VR technologies can contribute to improvement of the manufacturing processes
themselves.
350 P. Buń et al.
References
1. Gorski, F.: Building virtual reality applications for engineering with knowledge-based
approach. Manag. Prod. Eng. Rev. 8(4), 64–73 (2017)
2. Martín-Gutiérrez, J., Mora, C.E., Añorbe-Díaz, B., González-Marrero, A.: Virtual
technologies trends in education. EURASIA J. Math. Sci. Technol. Educ. 13(2), 469–486
(2017)
3. Hamacher, A., et al.: Application of virtual, augmented, and mixed reality to urology. Int.
Neurourol. J. 20(3), 172–181 (2016)
4. Ye, J., Badiyani, S., Raja, V., Schlegel, T.: Applications of virtual reality in product design
evaluation. In: 12th International Conference on Human-Computer Interaction (HCI
International 2007), Beijing, China, 22-27 July 2007. Published in: HCI 2007 Proceedings
of the 12th International Conference on Human-Computer Interaction: Applications and
Services (Part 4), vol. 4553, pp. 1190–1199 (2007)
5. Żywicki, K., Zawadzki, P., Górski, F.: Virtual reality production training system in the scope
of intelligent factory. In: Advances in Intelligent Systems and Computing, vol. 637, pp. 450–
458. Springer, Cham (2018)
factory of the future. Int. J. Comput. Integr. Manuf. 30(1), 182–190 (2017)
7. Langley, A., Lawson, G., Hermawati, S., D’Cruz, M., Apold, J., Arlt, F., Mura, K.:
Establishing the usability of a virtual training system for assembly operations within the
automotive industry. Hum. Fact. Ergon. Manuf. Serv. Ind. 26(6), 667–679 (2016)
8. Berg, L.P., Vance, J.M.: Industry use of virtual reality in product design and manufacturing:
a survey. Virtual Reality 21(1), 1–17 (2017)
9. Fogliatto, F.S., da Silveira, G.J., Borenstein, D.: The mass customization decade: an updated
review of the literature. Int. J. Prod. Econ. 138(1), 14–25 (2012)
10. Gorski, F., Zawadzki, P., Hamrol, A.: Knowledge based engineering as a condition of
effective mass production of configurable products by design automation. J. Mach. Eng. 16
(4), 5–30 (2016)
11. Lopes, J.J., Varela, M.L.R., Trojanowska, J., Machado, J.: Production flow improvement in
a textile industry. In: Rocha, Á., Guarda, T. (eds.) Proceedings of the International
Conference on Information Technology & Systems (ICITS 2018). Advances in Intelligent
Systems and Computing, vol. 721, pp. 224–233. Springer, Cham (2018)
12. Womack, J.P., Jones, D.T., Roos, D.: The Machine that Changed the World: The Story of
Lean Production. Rawson Associates, New York (1990)
S., Jurczyk, M. (eds.) Advances in Manufacturing. Lecture Notes in Mechanical
Engineering, pp. 131–140. Springer, Cham (2018)
14. Liker, J.K.: Becoming Lean: Inside Stories of U. S. Manufacturers. Productivity Press,
Portland (1997)
15. Womack, J.P., Jones, D.T.: Lean Thinking: Banish Waste and Create Wealth in Your
Corporation. Simon & Schuster, New York (1996)
16. Buń, P., et al.: Educational simulation of medical ultrasound examination. Procedia Comput.
Sci. 75, 186–194 (2015)
17. Buń, P., et al.: Possibilities and determinants of using low-cost devices in virtual education
applications. Eurasia J. Math. Sci. Technol. Educ. 13(2), 381–394 (2017)
18. Buń, P., Górski, F., Lisek, N.: Comparison of manual assembly training possibilities in
various virtual reality systems. In: International Conference on Innovation, Engineering and
Entrepreneurship, pp. 398–404. Springer, Cham (2018)
19. Górski, F., et al.: Effective design of educational virtual reality applications for medicine
using knowledge-engineering techniques. Eurasia J. Math. Sci. Technol. Educ. 13(2), 395–
416 (2017)
20. Hamblin, C.J.: Transfer of training from virtual reality environments. Thesis (Ph.D.).
Wichita State University (2005)
21. Trojanowska, J., Buń, P., Ivanov, V, Pavlenko, I.: The use of virtual reality training
application to increase the effectiveness of workshops in the field of lean manufacturing. In:
Proceedings of the International Conference of the Virtual and Augmented Reality in
Education (2018)
22. Rewers, P., Górski, F., Buń, P.: Augmented reality in production management classes. In:
Proceedings of the International Conference of the Virtual and Augmented Reality in
Education (2018)
A Case Study of a Virtual Training
Environment
Vasiliki Liagkou(&) and Chrysostomos Stylios
Department of Informatics and Telecommunications, University of Ioannina,

Kostakioi Arta, 47100 Ioannina, Greece
liagkou@kic.teiep.gr
Abstract. Virtual Reality is the most promising technology for providing new
learning and education solutions and more efficient opportunities for a revolu-
tionary training environment. Here we introduce a Learning Management Sys-
tem (LMS) for providing users not only with all education material but also
supporting experiential learning as well through VR simulations. The presented
VR model trains users for managing several threat factors, ensuring data pro-
tection and security for the operation of LPWAN in Industry 4.0 environment.
The training platform uses virtual reality environment so that to increase user’s
comprehension of the provided security level of an LPWAN network. In
addition to this, users of Virtual Reality training platform can easily discover the
security vulnerabilities of the LPWAN deployment. The proposed VR model
provides a web based portal that is incorporated with Moodle and OpenSim for
providing a user interface for managing Learning activities and for consistent
linking of learning activities in e-Learning and OpenSim environments.
Keywords: Virtual reality Learning Management Systems VR simulation

VR training LPWAN
1 Introduction
Over the last years the widespread usage of Internet and accessibility at educational
resources; the new technologies and various accessing means have led to the creation of
vast amount of open online educational environments. There have been examined and
proposed new concepts and methodologies towards the realization of e-participation of
students and at the same time aiming to the integral combination of academic and
industrial sector [1–6]. Many scientists work on the technologies and methods of
adaptive learning (see [7–12]). On the other hand the adoption of new technologies like
Virtual Reality in education could offer different more adaptive and attractive learning
experience. The development of 3-dimensional Virtual Worlds is a key element in
distance learning [13–15] offering a 3-D experience to users interaction and commu-
nication [16].
Virtual Reality (VR) has the ability to create and integrate any kind of environment,
redesign, retest and refined them in a virtual computer based framework. Thus, Virtual
Reality is the most promising technology for providing new learning and education
solutions and more efficient opportunities for a revolutionary training environment. The

https://doi.org/10.1007/978-3-030-18715-6_30
A Case Study of a Virtual Training Environment 353
use of Virtual reality for training and learning proposes [17–21] is attracted great
interest from the research community. Thus various VR use case scenarios are pro-
posed to learning users specific activities [22–26]. Beyond that, VR simulation on a
real-time event in industry can investigate poorly understood phenomena and allows
companies to decrease design and production costs [17, 18, 21, 23, 27–35]. In one of
those scenarios, they created a platform that simulates an actual facility. All buildings
and objects were created with the use of 3D CAD. The user with the use of HMD and
haptics could perform machining operations. The creation of a method for a risk
evaluation is critical for all industries. The VR has been used for evaluation and testing
[21, 28]. Furthermore VR simulation of a real-time movement decreases the potential
cost and risk of physical implementation thus Virtual Reality is undeniable a unique
tool that has many uses [36], from teaching students all around the world to simulate
safety scenarios [20, 37, 38]. It is a great substitute for the real life environment and can
minimize the safety concerns especially for training users to a new technology. Here
our VR model trains users to a new IoT technology.
Here we introduce a Learning Management System (LMS) for providing users not
only with all education material but also supporting experiential learning as well
through VR simulations. Moreover we implemented a simple user role playing 3D
game for delivering Information Security issues in LPWAN networks thought the LMS
platform that includes all the learning material for exploiting security parameters in
LPWAN. Our proposed training system includes an internet portal that provides to the
users an Learning Management System (LMS) that will support the exchange of
information on learning activities and uses the possibilities of the 3D virtual world
technology to create experiential learning simulations. Finally we evaluate the usage of
the developed VR training system by a set of 10 students.
1.1 VR Software Platforms

There are various software application that could be used to create Virtual Environment
with specific scenarios usable for training users not only in education sector but in
industry sector [39]. Unreal Engine 4 and Unity 3D are VR tools that can be used to
develop VR applications for training users in various scenarios but require heavy
coding. There is also the solution of Second Life and OpenSim. Those two platforms
have been used for years to create virtual worlds for education and learning. Both
OpenSim and Second Life are applications that are easy to install and maintain [40].
Unlike those two platforms Linden Lab, the company behind Second Life, they pub-
lished a new platform named Sansar. In this platform they upgraded the graphics to
provide improved immersion but it also raised the PC requirements. The realistic look
of a VR application makes the immersion better. Software such as blender and 3DS
max has been used to create 3D models. In the presented VR model we use the
OpenSimulator because it provides less immersion compared to platforms that have
been created from Unity 3D and Unreal Engine 4, it is widely used for educational
purposes and in some cases replace the classroom.
354 V. Liagkou and C. Stylios
1.2 LPWANs
There is a wide applied approach of exploiting sub-GHz bands in order to increase the
transmission range of nodes by trading-off data transmission rate while keeping at the
same time power consumption at low levels [41]. This so-called Low-Power Wide
Area Networks (LPWANs) are in contrast to high-frequency communication. In fact,
low frequency signals are not as attenuated by obstacles, thick walls or multipath
propagation and they are contributing in this way to robustness and reliability of the
signal [42].
LPWAN technologies allow IoT devices to connect to gateways (also called col-
lectors or concentrators) over distances in the range of several kilometers. Overall,
LPWANs are promising candidates for IoT applications in Industry 4.0, since they
allow high energy autonomy of the connected devices, low device and deployment
costs, high coverage capabilities and support a large number of devices [43].
1.3 Security in LPWANs

Security is a major concern for novel Industrial IoT environments, since primitively
industrial systems are not designed to deal into an internet not secure connected
environment. One major concern about the security threats in industrial IoT ecosystems
is the possibility to disclosure critical data and information. Exposure of critical data is
essential such as revealing the design of a new product, which could cause substantial
financial and reputation loss. Industrial competitors spend a lot of effort to gain some
inside information about products, suppliers, production procedures and methodolo-
gies. The disclosure of critical business data may allow access to the critical know-how
of a company and might result in a very high business risk.
Regardless of the technology used to connect the smart devices with the Internet in
the industrial environment, during the interaction within the IoT ecosystem, huge
amounts of data are being recorded, shared, aggregated, annotated, stored and pro-
cessed. If an external have access to these collected data may use it to extract or infer
sensitive information about the industrial process and so violate company’s and cus-
tomers’ privacy [21, 44].
It is essential that any IoT system must guarantee the confidentiality and integrity of
the information and the privacy and anonymity especially for industrial users. More-
over, it is important to guarantee the confidentiality of the end nodes data, especially
when these end node devices operate in open and uncontrolled environments. Inher-
ently, IoT framework suffers from several security issues which are more challenging
than any other case because of the complex and resources-constrained IoT devices
environment. There are initiated a lot of research efforts in order to investigate and
provide efficient security solutions for IoT environment, particularly to address
resources constraints and scalability issues. A comprehensive top-down survey of the
most recent proposed security and privacy solutions in IoT is provided in [45].
Recently some technical papers have concentrated on the security vulnerabilities in
LPWANs [46–48] providing alternative solutions for the cryptographic primitives (see
[49] and [50]). Authors in [10] focus on application server vulnerabilities and in [51]
introduced an alternative key management scheme. However, to the best of our
knowledge, there are no approaches that analyze the trust from a technological, policy
and public awareness point of view, based on a trust life cycle of a system that includes
the operation phases.
1.4 Our VR Model

In the present study is designed and implemented a Learning Management System
(LMS) that supports Virtual Reality technology for training purposes. The designed
system includes an LMS platform includes various learning and assessment activities
and implements educational activities in Virtual Reality environments. The imple-
mented system utilizes a Virtual training model that helps users to investigate the
security issues and tools in LPWAN protocols and to better understand new risks and
vulnerabilities that they should consider and address when they use LPWAN protocol.
The presented VR model utilizes the designed use case scenarios that were intro-
duced in [36, 52] for training users for managing several threat factors, ensuring data
protection and security for the operation of LPWAN in Industry 4.0 environment. The
training platform uses virtual reality environment so that to increase user’s compre-
hension of the provided security level of an LPWAN network. In addition to this, users
of Virtual Reality training platform can easily discover the security vulnerabilities of
the LPWAN deployment. The proposed VR model provides a web based portal that is
incorporated with Moodle, Sloodle and OpenSim for providing a user interface for
managing Learning activities and for consistent linking of learning activities in e-
Learning and OpenSim environments. Finally we evaluated the usage of the developed
VR training system by a set of 10 students.
1.5 Structure
The rest of the paper is organized as follows. Section 2 presents system’s architecture
and its components. Section 3 presents the unifying training environment and Sect. 4
summarizes its usage evaluation by a small set of users. The paper concludes in Sec. 5
along directions for future work.
2 System’s Architecture
The architecture of VR training system is consisted on the following main components:

• Link to the home page
• Template of courses
• Learning Activities
• Learning Assessment Tasks
• The Forum
• Help links to users’ manual, training course, and video on the use of the portal
• Professors’ and Students’ user interface
• The VR training Environment.
System’s architecture is consisted of various components that have different and

complementary functionalities and roles. In the following, we will give a detailed
description of the portal:
Home Page: This is the main informational web page. Through this web page, the
Users are informed about the system’s functionality and how it operates. It provides to
the users the necessary links to the other components of the system. Every time a user
desires to interact with the system, the first action to take is to visit the url of the portal
and then he can perform various actions (e.g. evaluate a competence, browse a Course
etc.). Home Page is accessible through http://www.tiphys.eu/moodle/ (see Fig. 1).
Fig. 1. Home page.
Professors Interface: A professor could log into the portal and access the interface in
order to:
• Edit/Insert data in a specific Course
• View a Course
• Add Announcements
• Insert/edit an Learning Activity
• Edit an VR environment of the uploaded learning activities and assessment tasks
Student Interface: A student could log into the portal and access the interface in order
to:
• View a Course
• View Announcements
• Access the list of Learning Activities/Assessment Tasks
• Participate in VR simulation for a of Learning Activity or Assessment Task
• View a Virtual Presentation
• Participate to a virtual exam
List of Courses: This component includes the index of all the provided Courses and
their links. Students could browse part or/all the provided Courses through this web
page.
Course Components
Each Course component consists of the following subcomponents:
1. Course’s Description: This component includes a detailed description for each
Course.
2. Announcements: This component includes and presents all the uploaded
announcements from professors for the specific Course.
3. Teaching Learning Activities: This component describes the Teaching Learning
Activities for each Course for learning Industry 4.0.
4. Assessment Tasks: This component helps students to understand what actual
assessment tasks.
5. Must be delivered by a specific Course. Professors are responsible for constructing
these tables.
6. VR simulation: This component is responsible for the realization of virtual training
environment for a specific assessment task or learning activity (see Fig. 2).
7. VR components: This component uses VR representation for participating a VR
class attendance or a VR exam (see Fig. 3).
The developed system includes an open LMS website. The LMS was developed using
the Moodle technology. Moodle is installed in a Linux server that operates Linux
Ubuntu 16.04.4 and it uses mysqli as a database. Moodle provide the users with tools to
create their own courses, communicate via messages with other users and post ques-
tions in dedicated forums for each course.
Fig. 2. VR simulation.
Each course includes a list of Learning Activities and Assessment tasks and some of
them are connected with the VR training environment.
Fig. 3. VR presentation and VR exam.
3 The VR Training Environment
Virtual Reality training environment operates via OpenSimulator server and user’s
client uses the Firestorm viewer. The LMS system is connected with the VR training
environment via the HOP technology. The learning activities and assessment tasks
(Fig. 1) are connected with a set of URLs in the virtual reality server and when users
from the browser click on the link he will be teleported to the specific locations of
training (Fig. 2).
The creation of the included 3D objects in the implemented VR training environ-
ment was made via OpenSimulator’s native tool and via creating collada files by using
blender or any other software for 3D object creation.
The developed scripts in VR Training environment are created in server by using
the Linden Script Language for short LSL. LSL is a native scripted language developed
and used in the OpenSimulator server.
The users in the VR Training Environment can navigate in Virtual World and
interact with it. They can view lectures, do exams and participate in quiz questions.
They can also use the chat function or the voice chat to communicate with others.
VR presentation was created as an object using the OpenSimulator’s native tools.
VR presentations are uploaded in server, by creating an image for each slide, and
imported them in its object. The developed scripts in the LSL language provide VR
view function of presentation’s object.
The VR exams and quizzes were created as a small object and it is added a small
image. A notecard with the questions was created and imported to their object. The
developed scripts in the LSL language provide a VR read function of the notecard and a
small dialog box that opens on the top right side of the user for answering them. All the
answers are recorded and can be seen by other users in the chat.
We implemented a “Security in LPWAN” course in the presented training envi-
ronment for evaluating user’s comprehension of the security parameters of an LPWAN
network by a small set of ten students. The presented training environment implements
a 3D game where the specific tools and approaches ensuring that users know the basic
security tools of LPWAN protocol and understand the potential security flaws for their
actions. Therefore the course “Security in LPWAN” includes the following set of
learning activities and assessment tasks that are linked to a VR training environment
where the user has to pass through different rooms/areas for concluding a 3D game.
The course “Security in LPWAN” includes the following list of Learning Activities
and Assessment Tasks that are connected with the VR training environment:
1. LPWAN Network Architecture
2. Keys’ Security Parameters
3. How does the Joining Procedure work
4. How to send a packet
In all implemented VR environments user’s avatar could make specific actions like
navigation, interaction with objects, communication, objects creation and avatars
configuration within the virtual world. The main 3D experience of the Learning
activities and assessment tasks and how it helped the end-users are described in fol-
lowing subsections.
3.1 LPWAN Network Architecture

The designed training environment includes specific visualizations and steps so that the
user to be able to understand the basic security parameters of LPWAN network.
Initially users navigate thought a LPWAN network where the avatar can select the
connection links from different components as shown in Fig. 4. One of the basic
Fig. 4. LPWAN components.

characteristics that LPWAN introduces in order to perform all its provided security
mechanisms is that the user interacts through various concentrators to an application
server that forwards the encrypted package to Network Server. The 90% of all end-
users stated that 3D virtual environment helped them to understand LPWAN compo-
nents and their functionality.
The assessment task for this learning activity includes a 3D game where the Avatar
has to select and connect an IoT device and thought its navigation is able to assess the
knowledge about the basic security factors of LPWAN by answering specific questions
in different parts of its navigation based on the acquired Virtual experience it possess.
3.2 How Does the Joining Procedure Work?

Here the VR environments simulates the learning activity by informing the user about
LPWAN joining procedures via a VR presentation as shown in Fig. 3. The user after
concluding the presentation he must be able to understand that the LPWAN network
provides two different procedures for connecting an IoT device with different security
requirements. All the users stated that the VR presentation has minor benefits compared
with the uploaded 2D presentation on LMS portal.
The assessment task for this learning activity is utilized by a quiz (see Fig. 5).
Based on user’s answers the user could follow two different joining scenarios. Here the
majority of users do not understood all the security parameters of the joining
procedures.
Fig. 5. Knowledge assessment
3.3 Keys’ Security Parameters

In this VR experience the user will learn how the corresponding keys are used by
constructing package information, encrypting and signing it.
This VR experience implements the learning activity by a more experiential way

and users stated that got deeper knowledge on this environment compared to the
previous ones.
The packet construction VR experience helps the user to understand how an IoT
device is connected in LPWAN. User’s avatar can create its packet by selecting
packet’s component by a stack of objects. The packet construction VR experience helps
the user to understand what information is broadcasting by LPWAN protocol for
joining the network. Figure 6 shows the packet construction phase and how it helps the
avatar to be informed:
• for its actions.
• that the initial secret symmetric is known to both end-device and the network server.
• that the IoT device is equipped with an application identifier and a global unique
identifier.
Fig. 6. Packet construction
The avatar could also encrypt its constructed packet by throwing a dice for generating a
random number and by locking the packets by using a secret key for encrypting the
transmitted packet (see Figs. 7 and 8).
Finally the assessment task for this learning activity includes several quizzes that
are placed on different scenes of the VR environment (Fig. 9). The majority of the end
users stated that this VR experience helped them to increase their comprehension about
how the provided keys are used.
3.4 How to Send a Packet

The VR environment that implements this learning activity simulates the generation of
all the basic keys of LPWAN and visualizes where they are used. In this VR scenario
the avatar generates its secret keys by using an 128 bit key and the user has to select at
Fig. 7. Generating a nonce
Fig. 8. Encrypting and sign
a panel either the Network Secret Key or Application Secret Key (Fig. 10). Moreover
this training phase visualize when the encryption is performed with the Application
Secret Key or the Network Secret Session Key and inform the user about the usage of
Application Secret Key and Network Secret Session Key and their relation to packet’s
content.
Fig. 9. Assessment quizzes
Fig. 10. Getting secret keys
The assessment task checks user’s knowledge about the use of the generated ses-
sion keys (Application Secret Key and Network Secret Session Key) for encrypting all
further message payloads between the node and the application server during the
lifetime of the session (Fig. 11). The 80% present of users stated that this VR envi-
ronment mainly was helped them to understand how the application server could not be
able to decrypt their packet.
Fig. 11. Sending a packet
4 End-Users Results
The most important result obtained from the evaluation of our VR training environment
from all of ten end users is the general perception that a training system that provides
all the education material and also supports VR learning simulations provides a very
convenient training experience. The majority of users are confident that 3D virtual
worlds can provide learners with a full understanding of a situation using experiential
3D use case scenarios.
Users in general prefer a VR use case scenario that simulates several real life
actions in order to obtain the sense of purpose of the simulating action, thus the VR use
case scenarios that only include 3D navigations and quizzes do not provide good
conceptual understanding.
Thus use case scenarios that simulates the learning activities must include several
interactions with virtual objects in order to enhance learners’ interest and engagement
to the learning tasks and help them to develop a stronger conceptual understanding.
5 Future Work
Here we proposed a VR training system that includes an internet portal that provides to
the users an Learning Management System (LMS) that will support the exchange of
information on learning activities and uses the possibilities of the 3D virtual world
technology to create experiential learning simulations. Moreover we implemented a
simple user role playing 3D game for delivering Information Security issues in
LPWAN networks thought the LMS platform that includes all the learning material for
exploiting security parameters in LPWAN and we evaluated its usage to a small set of
users.
Although Sloodle (Simulation Linked Object Oriented Dynamic Learning Envi-
ronment) provides various tools for integrating the Moodle with Second Life it is
outdated and all the virtual tools except presenter were encountered various compati-
bility problems and errors. We plan to evaluate the usage of the developed VR training
system by using a bigger set of students in Computer Security theme. Based on the
evaluation, we will further update the environment so that to be able to successfully
educate the interested parties.
Acknowledgments. This work has been partially supported by the “TIPHYS 4.0 - Social
Network based doctoral Education on Industry 4.0” project No 2017-1-SE01-KA203-03452
funded by ERASMUS+ of the European Commission.
References
1. Berge, Z.L.: The role of the online instructor/facilitator. Educ. Technol. 35(1), 22–30 (1995)
2. Bule, J., et al.: Expert model development. In: Proceedings of the 10th IASTED International
Conference on Computers and Advanced Technology in Education, Beijing, China, 8–10
October 2007, pp. 81–84 (2007)
3. Bule, J., et al.: E-course development based on the model “system assembly from reusable
component”. In: Proceedings of the 8th International Conference on e-Learning (ICEL
2013), Cape Town, South Africa, 27–28 June 2013, pp. 69–76 (2013)
4. Jones, E.A., et al.: Defining and assessing learning: exploring competency-based initiative.
Institute of Education Sciences, National Center for Education Statistics, 6 November 2002.
http://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2002159
5. Klein-Collins, R.: Competency-based degree programs in the US: postsecondary credentials
for measurable student learning and performance, Council for Adult and Experiential
Learning (2012). www.cael.org/pdfs/2012_ competencybasedprograms
6. Klein, J.D., et al.: Instructor competencies: standards for face-to-face, online, and blended
settings. http://www.tutzauer.com/TLC/Teaching_competencies.pdf. Accessed 15 May 2013
7. Liagkou, V., Stylios, C.: An E-competence evaluation portal for software engineering master
course. Int. J. Inf. Technol. Secur. 9, 27–38 (2017). ISSN 1313-8251
8. Liagkou, V., Stylios, C.: The case study of a software engineering e-competence evaluation
portal. In: Actual Problem of Education, pp. 21–23 (2017)
9. Maureen, E., et al.: Representing targets of measurement within evidence-centered design.
Appl. Meas. Educ. 23(4), 325–341 (2010)
10. Michorius, J.: Whats mine is not yours: Lora network and privacy of data on publishing
devices. In: 25th Twente Student Conference on IT (2016)
11. Misnevs, B.: E-learning in Latvia, dimensions of e-learning education. In: Proceedings of the
Cases on Challenges Facing E-Learning and National Development. Institutional Studies
and Practices, Latvia, pp. 379–409 (2010)
12. Misnevs, B., et al.: Data science: professional requirements and competence evaluation.
Baltic J. Mod. Comput. Riga 3(4), 441–453 (2016)
13. De Freitas, S.: Learning in immersive worlds: a review of game-based learning. In: Joint
Information Systems Committee, Bristol (2006)
14. De Freitas, S.: Serious virtual worlds report. JISC, Bristol (2008). http://www.jisc.ac.uk/
publications/reports/2008/seriousvirtualworldsreport.aspx
15. Dickey, M.D.: Three-dimensional virtual worlds and distance learning: two case studies of
active worlds as a medium for distance education. Br. J. Educ. Technol. 36, 439–451 (2005)
16. Kahiigi, E.K., Ekenberg, L., Hansson, H., Tusubira, F.F., Danielson, M.: Exploring the e-
learning state of art. Electron. J. e-Learn. 6, 77–88 (2008)
17. Makarova, I., Khabibullin, R., Belyaev, E., Bogateeva, A.: The application of virtual reality
technologies in engineering education for the automotive industry. In: International
Conference on Interactive Collaborative Learning, Florence, pp. 536–544 (2015)
18. Matsas, E., Vosniakos, G.C., Batras, D.: Prototyping proactive and adaptive techniques for
human-robot collaboration in manufacturing using virtual reality. In: Robotics and
Computer-Integrated Manufacturing, pp. 168–180 (2018)
19. Němec, M., Fasuga, R., Trubač, J., Kratochvíl, J.: Using virtual reality in education. In: 15th
International Conference on Emerging eLearning Technologies and Applications, Stary
Smokovec, pp. 1–6 (2017)
20. Parlangeli, O., Palmieri, S., Mariani, M., Tartaglia, R., Arcangeli, G., Bagnara, S.: Usability
issues when applying virtual reality in education and training: a case study. In: Proceedings
of the Human Factors and Ergonomics Society Annual Meeting, pp. 628–631 (2000)
21. Price, B.A., Adam, K., Nuseibeh, B.: Keeping ubiquitous computing to yourself: a practical
model for user control of privacy. Int. J. Hum Comput Stud. 63(1), 228–253 (2005)
22. Carruth, D.W.: Virtual reality for education and workforce training. In: 15th International
Conference on Emerging eLearning Technologies and Applications, Stary Smokovec, pp. 1–
6 (2017)
23. Hamid, N.S.S., Aziz, F.A., Azizi, A.: Virtual reality applications in manufacturing system.
In: Science and Information Conference, London, pp. 1034–1037 (2014)
24. Seth, A., Vance, J.M., Oliver, J.H.: Virtual reality for assembly methods prototyping—a
review. In: Virtual Reality, pp. 5–20. Springer, Cham (2011)
25. Stuchlíková, L., Kósa, A., Benko, P., Juhász, P.: Virtual reality vs. reality in engineering
education. In: 15th International Conference on Emerging eLearning Technologies and
Applications, Stary Smokovec, pp. 1–6 (2017)
26. Zhang, K., Suo, J., Chen, J., Liu, X., Gao, L.: Design and implementation of fire safety
education system on campus based on virtual reality technology. In: Federated Conference
on Computer Science and Information Systems, pp. 1297–1300 (2017)
27. Berg, L.P., Vance, J.M.: Industry use of virtual reality in product design and manufacturing:
a survey. In: Virtual Reality, Springer, London, pp. 1–17 (2017)
28. Büttner, S., Mucha, H., Funk, M., Kosch, T., Aehnelt, M., Robert, S., Röcker, C.: The design
space of augmented and virtual reality applications for assistive environments in
manufacturing: a visual approach. In: International Conference on PErvasive Technologies
Related to Assistive Environments, pp. 433–440 (2017)
29. Calvo, I., López, F., Zulueta, E., González-Nalda, P.: Towards a methodology to build
virtual reality manufacturing systems based on free open software technologies. Int.
J. Interact. Des. Manuf. 569–580 (2017). Springer, Paris
30. Gamlin, A., Breedon, P., Medjdoub, B.: Immersive virtual reality deployment in a lean
manufacturing environment. In: International Conference on Interactive Technologies and
Games, Nottingham, pp. 51–58 (2014)
31. Gandhi, R.D., Patel, D.S.: Virtual reality – opportunities and challenges. In: International
Research Journal of Engineering and Technology, pp. 482–490 (2018)
32. Choi, S., Jung, K., Noh, S.D.: Virtual reality applications in manufacturing industries: past
research, present findings, and future directions. In: Concurrent Engineering, pp. 40–63
(2015)
33. Darmoul, S., Abidi, M.H., Ahmad, A., Al-Ahmari, A.M., Darwish, S.M., Hussein, H.M.A.:
Virtual reality for manufacturing: a robotic cell case study. In: 2015 International Conference
on Industrial Engineering and Operations Management, Dubai, pp. 1–7 (2015)
34. Hutabarat, W., Oyekan, J., Turner, C., Tiwari, A., Prajapat, N., Gan, X.P., Waller, A.:
Combining virtual reality enabled simulation with 3D scanning technologies towards smart
manufacturing. In: Winter Simulation Conference. Washington, pp. 2774–2785 (2016)
35. McGrath, J.L., Taekman, J.M., Dev, P., Danforth, D.R., Mohan, D., Kman, N., Crichlow,
A., Bond, W.F.: Using virtual reality simulation environments to assess competence for
emergency medicine learners. Acad. Emerg. Med. 25, 186–195 (2018)
36. Liagkou, V., Salmas, D., Stylios, C.: VR training model for exploiting security in LPWAN.
In: Conference on Intelligent Computation in Manufacturing Engineering, Gulf of Naples,
Italy (2018)
37. Gavish, N., Gutiérrez, T., Webel, S., Rodríguez, J., Peveri, M., Bockholt, U., Tecchia, F.:
Evaluating virtual reality and augmented reality training for industrial maintenance and
assembly tasks. In: Interactive Learning Environments, pp. 778–798 (2013)
38. Li, X., Yi, W., Chi, H., Wang, X., Chan, A.P.: A critical review of virtual and augmented
reality (VR/AR) applications in construction safety. Autom. Constr. 86, 150–162 (2018)
39. Liagkou, V., Salmas, D., Stylios, C.: Realizing virtual reality for Industry 4.0. In: CIRP
Conference on Intelligent Computation in Manufacturing Engineering, Gulf of Naples, Italy
(2018)
40. Christopoulos, A., Conrad, M.: Maintaining context in a changing (virtual) world -
educators’ perspectives for OpenSim and second life. In: 5th International Conference on
Computer Supported Education, Aache, Germany (2013)
41. Centenaro, M., Vangelista, L., Zanella, A., Zorzi, M.: Long-range communications,
unlicensed bands: the rising stars in the IoT and smart city scenarios. IEEE Wirel. Commun.
23(5), 60–67 (2016). https://doi.org/10.1109/MWC.2016.7721743
42. Raza, U., Kulkarni, P., Sooriyabandara, M.: Low power wide area networks: an overview.
IEEE Commun. Surv. Tutor. 19, 855–873 (2017)
43. Boulogeorgos, A., Diamantoulakis, P., Karagiannidis, G.: Low power wide area networks
(LPWANs) for internet of things (IoT) applications: research challenges and future trends.
CoRR, abs/1611.07449 (2016)
44. Pavlou, P.: State of the information privacy literature: where are we now and where should
we go? MIS Q. 35(4), 977–988 (2011)
45. Kouicem, D.E., Bouabdallah, A., Lakhlef, H.: Internet of things security: a top-down survey.
Comput. Netw. 141, 199–221 (2018)
46. Aras, E., Ramachandran, G.S., Lawrence, P., Hughes, D.: Exploring the security
vulnerabilities of Lora. In: IEEE International Conference on Cybernetics (CYBCONF
2017), pp. 1–6 (2017)
47. Chatzigiannakis, Y., Liagkou, V., Spirakis, P.: Providing end-to-end secure communication
in low-power wide area networks (LPWANs). In: International Symposium on Cyber
Security Cryptography and Machine Learning (CSCML 2018) (2018)
48. Emekcan, A., Gowri, R., Piers, L., Danny, H.: Exploring the security vulnerabilities of Lora.
In: IEEE International Conference on Cybernetics (CYBCONF), pp. 1–6 (2017). https://doi.
org/10.1109/cybconf.2017.7985777. Accessed June 2017
49. Kim, J., Song, J.: A simple and efficient replay attack prevention scheme for LPWAN. In:
International Conference on Communication and Network Security (ICCNS 2017), pp. 32–
36 (2017). http://doi.acm.org/10.1145/3163058.3163064
50. Miller, R.: Lora security: building a secure Lora solution. Technical report, MWR Labs
(2016)
51. Naoui, S., Elhdhili, M.E., Saidane, L.A.: Enhancing the security of the IoT LPWAN
architecture. In: International Conference on Performance Evaluation and Modeling in Wired
and Wireless Networks (PEMWN 2016), pp. 1–7 (2016)
52. Liagkou, V., Stylios, C.: A trustworthy and privacy preserving model for online competence
evaluation system. In: International Conference on Dependability and Complex Systems.
AISC, vol. 761, pp. 338–347. Springer, Cham (2018). ISBN 978-3-319-91445-9
Virtual and Augmented Reality
for Configuring, Promoting and Servicing
Household Appliances
Krzysztof Walczak1(&), Jakub Flotyński1, Dominik Strugała1,

Dariusz Rumiński1, Mikołaj Maik1, Anna Englert1,
and Tomasz Jenek2
1
Poznań University of Economics and Business,
Niepodległości 10, 61-875 Poznań, Poland
walczak@kti.ue.poznan.pl
2
Amica S.A., Mickiewicza 52, 64-510 Wronki, Poland
Abstract. In this paper, we present several virtual and augmented reality sys-
tems for interactive visualization of household appliances. The presented sys-
tems can support the wholesale of series of products configured and then ordered
in a manufacturers’ showroom as well as individual sales by enabling customers
to visualize different versions of the products in their home setting before
making a purchase decision. The system provides also training and support for
servicing of household appliances. The visualization software, created with the
use of the Unity3D engine, takes advantage of a wide range of visualization
hardware, such as a smartphone, an HMD, and a Powerwall, to enable its use in
different settings. With the use of the developed solutions, users can configure
and manipulate components of virtual home appliances directly in a 3D space.
The visualization software, which has been developed in cooperation with
Amica S.A. – a major household appliance manufacturer in Poland, can be an
important marketing element and an innovation that has the potential to attract
new customers.
Keywords: Virtual reality Augmented reality Visualization HMD

Household appliances
1 Introduction
Presentation and simulation of products in the manufacturing sector can largely benefit
from the introduction of advanced IT solutions. Virtual reality (VR) and augmented
reality (AR) technologies offer significant advantages due to their potential to immerse
users in interactive computer-generated synthetic worlds. Fast prototyping, marketing
and servicing are the areas in which the use of VR/AR technologies provides clear
benefits. In this work, we show how these technologies can be used to support the sales
of products at the wholesale and individual levels, as well as servicing of household
equipment.

https://doi.org/10.1007/978-3-030-18715-6_31
VR and AR for Configuring, Promoting and Servicing Household Appliances 369
The use of VR/AR in the manufacturing field represents a topic of growing interest
for both researchers and companies [1]. The potential of VR/AR applications can be
exploited also to promote new products, thus influencing customers’ decisions. The use
of advanced technologies may lead to an increased interest in a given brand of prod-
ucts. Being aware of this, companies are trying to enhance product visualization, by
introducing VR/AR/3D visualization techniques in their marketing campaigns. In this
way, potential customers are given the possibility to experience a virtual space as a
counterpart to a real exhibition that may be located in a remote place. This may help
customers in making more conscious decisions and creating valid expectations on how
the real experience would be like [2]. These aspects can have a significant impact on
the popularity of new forms of marketing [3].
Recent studies of human perception are reporting that the common audience has
become more resistant towards traditional visual media as a source of information [4].
It was hypothesized that only information provided in an innovative way is able to
attract new customers. A solution to this problem may be the use of VR/AR for
manufacturing and marketing. This can be a direction towards innovation for many
companies.
Although virtual reality has been until now mostly used for effective visualization of
existing products or designs, more and more scientists try to use VR/AR technology also
for design decision-making studies. VR/AR has a great potential as a marketing research
tool for understanding customers behavior and preferences. Implementation of the
possibilities of device customization allows to automatically gather data by tracking
users’ behavior. The obtained data about market trends can be analyzed, therefore
manufacturers can supply products in response to forecast demand. Given the rapid
changes that take place on the market, companies need to use information technologies
to improve market detection capabilities to simplify operational factors, knowing that
the process of launching a new product involves complex research and development [5].
In this paper, we demonstrate how VR, AR and 3D web technologies can be used
for configuration, visualization, marketing and servicing of household equipment. The
tools were developed for Amica S.A – a major household equipment manufacturer in
Poland.
The remainder of this paper is organized as follows. First, an overview of the
related works is provided. Then, several prototypes of the developed systems are
presented – the virtual product configurator, 3D web visualization of products, AR and
VR visualizations of products, and an immersive VR service guide. Finally, conclu-
sions are provided and future works are discussed.
2 Related Works
Creating practical interactive VR applications is a complicated task. This is due to the

conceptual and the structural complexity of VR content models and the diversity
aspects that must be taken into account in the content creation process. To build an
easy-to-use, configurable VR system, it is important to choose appropriate methods of
370 K. Walczak et al.
user interaction with the virtual environment. It is necessary that the interaction
methods are intuitive and user-friendly. New users often have difficulty navigating and
interacting in VR environments.
2.1 3D Modeling and Interaction

Geometry, appearance and movement of real objects can be acquired using automatic
or semi-automatic 3D scanning and motion capture devices. Static 3D objects can be
precisely digitized using active scanners based on laser ToF measurement, triangula-
tion, and structured light. Less precise, but more affordable, are software tools enabling
reconstruction of 3D objects from series of images, such as Autodesk 123D and
3DSOM. 3D scanning can be combined with other content creation methods, allowing
designers to influence the digitization process and the created content.
Modeling of both existing and non-existing objects is possible with the use of
visual 3D content design environments. Software packages that enable modeling or
sculpting 3D content include Blender, 3ds Max, Modo, Maya, ZBrush and 3D-Coat.
Advanced professional environments offer rich capabilities of modeling various content
elements, but their complexity requires high expertise. Narrowing the domain of
application and the set of available operations enables development of tools that are
easier to use by domain experts. Examples of such environments include AutoCAD
Civil 3D, Sweet Home 3D and Ghost Productions. These tools enable relatively quick
and efficient modeling without requiring users’ extensive experience in 3D content
creation. Such an approach, however, significantly reduces the generality of the content
creation process.
High structural complexity of 3D content, combined with the requirement of being
able to adjust specific content parameters, require the development of content models –
well-defined structures, which describe content organization and parameterization [6].
Based on such models, the final form of 3D content can be generated by content
generation software – either fully automatically or semi-automatically in an interactive
process. Content models offer data structures that are better organized and easier to
maintain than typical 3D content representation. They also permit automatic verifica-
tion of data consistency and elimination of redundancy. Content patterns provide an
additional conceptual layer on top of content models, defining roles of specific ele-
ments in the model [7, 8].
As an alternative to fixed content models, rules of content composition can be
defined. Such rules describe how different types of content elements should be com-
bined to form the final 3D model. Rules permit flexible composition of content from
predefined building blocks – components [9–11]. Components may represent geo-
metrical objects, scenarios, sounds, interaction elements, and others. Content creation
based on configuration of predefined components constrains possible forms of the final
created content. In many application domains, however, this approach is sufficient,
while the process is much simpler and more efficient than creating content from scratch.
To further simplify content modeling, separation of concerns between different

categories of users is required. These users may have different expertise and may be
equipped with different modeling tools. A non-expert designer may use ready-to-use
components and assemble them into virtual scenes. Composing a scene in such a way is
relatively simple, but the process is constrained. New content creation capabilities can
be introduced by programmers or 3D designers, who can add new components and new
ways of combining them.
The use of semantic web techniques may further simplify the process of creating
3D content [12–15]. Semantic web techniques enable the use of high-level domain-
specific concepts in the content creation process, instead of low-level concepts specific
to 3D graphics. Content creation may be supported by knowledge inference. The use of
semantic content representation enables creation of content that is platform indepen-
dent. Several approaches have been proposed to enable 3D content modeling with the
use of semantic web techniques [16–22].
Availability of efficient and easy to use content creation methods – in particular
methods based on the semantic web techniques – opens the possibility of social 3D
content co-creation by users who are both producers and consumers (prosumers),
similarly as in the case of the “two-dimensional” Web 2.0. 3D content sharing portals,
such as Unity Asset Store [23], Highend3D [24], Turbosquid [25], 3D ContentCentral
[26], and many others (e.g., CG People Network, Creative Crash, 3d Export, Ever-
motion, The 3D Studio, and 3D Ocean) enable access to vast libraries of 3D content.
However, the available content is not suitable for designing personalized products and
showrooms.
A vivid approach to user interaction in VR environments is the analysis of the
natural behavior of a human being. These techniques focus on providing an intuitive,
natural interface that is user-friendly even to non-experienced users. Using image
tracking techniques [27] or specialized sensor systems, such as the Xbox Kinect [28],
user movement can be captured. Gesture recognition [29], eye tracking [30] and voice
recognition [31] are also popular techniques.
2.2 VR and AR in Manufacturing and Marketing

There are vast possibilities to use VR to improve productiveness in manufacturing [32],
versioning [33], visualization [34], and marketing [35]. These fields can also influence
and extend one another using the capabilities of VR. In marketing, the applications of
virtual reality are diverse. For example, in the food industry, VR can be used to
strengthen brands, as Coca-Cola did in 2014 during the World Cup in Brazil [36].
Viewers were shown a VR presentation of a football match with users as players and
after taking some virtual physical effort they “earned” their cold can of drink. Ver-
sioning helps with lowering cost of creating multiple versions of products. One of such
applications is creating virtual environment, for configuring white goods [33]. In this
system, users can choose many components from a library and assemble them together
to the point where they are pleased with the result.
Interactive visualization of products in 3D can improve opinions about a brand – by

association with modern technologies and better communication with clients. Virtual
reality is also an important visualization technique in architecture, where presentation
of designs is the key element. People can better understand architectural spaces by
navigating and interacting within the environment.
3 Virtual Product Configurator
For a customer, product presentations are usually limited to showing a film, a mockup,
or a finished product. This inhibits any interaction and configuration of the presented
objects. Manufacturers should have greater freedom in demonstrating their products in
different variants without the need to create multi-format exhibition stands or to record
each product in all available configurations on videos. Nowadays, the ability of high-
quality visualization of objects and their subsequent presentation is one of the basic
marketing requirements. Presentations of this type should give an easy access to the
available elements (models uploaded to the application database, which are continually
updated), but also should give the ability to inspect items – rotate them, move or zoom
on individual elements, depending on what is currently necessary for a comprehensive
presentation. Typically, this level of interactivity in presentation is not achievable using
physical contact with real products, which is limited to touch and show only one of the
variants.
More and more frequently, customers demand highly personalized products, and
providing them is not always possible on a regular basis. Therefore, the possibility of
configuring elements of a given product using high-quality model visualization in 3D,
which facilitates the process of selecting preferred finishing parts of the product, is an
important innovation. The big advantage of this type of presentation is the ability to
almost instantly visualize the features that are difficult to demonstrate in reality, such as
the possibility of presenting particular components of the presented product or par-
ticular processes occurring in them – for example air circulation in an oven. These
types of elements are not only key components of a comprehensive visualization, but
are also essential for presentation of the functionality of a given appliance and for
creation of useful training materials.
In Fig. 1, an interactive 3D presentation of an oven on a Powerwall screen is
shown. The presentation is based on the use of the Unity3D environment, enabling the
creation of two- and three-dimensional interactive content. Unity3D is an engine,
which has been used for creating a multitude of computer games, and visualization in
many domains, e.g. architecture. The application shown in Fig. 1 was implemented
using the C# language, which enabled to code scripts providing extended interaction
with the product. A wireless Xbox controller is used to make navigation through the
presentation easier. 3D product models provided in the STEP format were used to
faithfully reproduce the products. This enabled transfer of precise models from the
design stage to visualization. Thanks to the use of a large high-resolution display, the
presentation can be shown to large groups at the same time, with an accurate
representation of even the smallest details. Size and quality of the displayed image
allows one to see elements that would not be visible on a normal monitor screen or in
the real product.
Fig. 1. An interactive animated virtual oven displayed on a Powerwall.
4 3D Web Visualization of Products
The ability to visualize 3D products over the web is a major requirement for marketing
and practical product presentation. Home users should be able to easily access and
visualize products directly from their web browsers, without the need to have access to
HMD devices or professional VR installations. 3D visualization should provide the
capabilities of rotating, moving and zooming the products depending on the user’s
needs. It should be possible to interact with the product itself – switch it on and off,
open, close, and move elements of the product. These features are not only important
for convincing visualization, but are critical for demonstration of product’s function-
ality and features.
In Fig. 2, an interactive 3D visualization of a product in a web browser is presented.
The visualization is based on the use of the X3D [37] standard for 3D content
description and X3DOM technology for presentation [38]. X3DOM is an open source
JavaScript framework, designed for presentation of declarative X3D scenes in web
browsers. X3DOM uses WebGL for efficient rendering of complex 3D content [39].
Due to the use of X3DOM, the 3D presentation can be displayed in all contemporary
web browsers without the need to install any additional plugins or external software.
Fig. 2. The virtual oven displayed in a web browser.
5 Visualization of Products Using AR
AR, with its potential to blend real and virtual objects, creates new opportunities for
promoting manufacturer’s products in an interactive way. Widespread use of AR has
been enabled in the recent years by remarkable progress in consumer-level hardware
performance, in particular, in the computational and graphical performance of com-
puters and quickly growing bandwidth of cellular networks. The progress is particu-
larly visible in the domain of mobile devices, such as smartphones, tablets, and
wearables. These types of devices are now regularly used by potential customers and
can be used to visualize household appliances in-place with the use of AR techniques.
Before making a decision on purchasing a product, a customer would like to check
whether the product matches the space among other appliances and furniture. Giving
customers an option of AR visualization of products before buying them could help
them making a final decision. For example, a customer’s kitchen surrounding can be
enriched with a 3D model representing an oven in order to check whether the product
matches the style and the size of the kitchen furniture.
Not only the size of the product is important, but also its design, color, and
functionality are significant factors that drive the customer to buy the product. If the
customer is satisfied with a model of a kitchen appliance, an AR application could also
provide a function to order the real product. Figure 3 depicts a virtual oven presented
on a marker in an AR environment using a mobile application, which has been
implemented for the Android operating system using the Vuforia library [40].
Fig. 3. A 3D model of an oven within an AR environment displayed on a smartphone.
6 Immersive Visualization of Products
To support immersive visualization of products we have developed a prototype system

enabling users to experience product models in a virtual showroom space. For this
purpose, we used an HTC Vive Pro device with two controllers. The main reason for
choosing HTC Vive Pro was that this HMD is equipped with relatively high-resolution
displays and has a system of continuous tracking without blind spots. This gives users
greater mobility when viewing the household appliances from different sides. In
addition, moving around the showroom by walking feels more natural. The system has
been implemented in Unity 2018.2.10f1 (64-bit) with the use of the C# language and is
running on a Windows 10 computer with the i7-8700 K (3.70 GHz) processor and the
GeForce GTX 980Ti graphics card.
The HMD changes the vision angle according to the movements of the user’s head.
A user can move freely around the environment and see all displayed items, making the
experience more comfortable and realistic (Fig. 4). All virtual household objects are
based on their original design models that were exported to the FBX format and then
transferred to Unity3D, where all related interactions were programmed.
Fig. 4. A virtual exhibition of ovens displayed in an HMD.

Navigation and interaction are implemented using the HTC Vive controllers, which
handle tracking of user’s hands enabling semi-natural interaction within the environ-
ment (e.g., opening oven door or interacting with equipment in its interior). These
devices provide comfortable navigation in the three-dimensional space, and in the
context of a virtual showroom they allow users to indicate particular household
appliances and areas of the showroom with a high level of precision.
Two movement modes are used to move around the scene. The first is teleportation,
which is often used in immersive VR applications. By aiming the left controller at the
ground and pressing the touchpad on top of it, the image of a viewfinder appears.
Releasing the button causes a user to teleport to the given place. This mode is default
and is intended to reduce the motion sickness effect. Some users have, however,
problems with moving around in a virtual space using this method. In such a case, they
can switch to the second navigation mode, where movement is triggered by pushing the
given direction on a touch pad. By pressing the trigger button on the right controller, a
user can grab and manipulate any object in the scene.
When a user wants to change configuration of an oven, a menu appears informing
the user about configurable parts of the product of choice. For example, a user may
select the type of display in the center of the main oven panel and the shape of knobs
located on the sides (Fig. 5a). In the next step, the user can choose the type of the door
handle (Fig. 5b). A user can quickly switch between configuration menu and presen-
tation mode throughout the entire scenario. New accessories are added to the config-
urator when they become available in production.
(a) (b)
Fig. 5. Configurator of a virtual oven.
7 Immersive Service Guide
An immersive service guide for an induction hob has been developed on the basis of an
HMD. The guide presents several scenarios about the repair of the most common
defects that can occur while using the appliance. In contrast to a typical paper guide,
the HMD enables intuitive user-friendly interactive presentation of the hob at a high
level of detail that is appropriate for step-by-step training how defects should be
repaired. The immersive guide permits to look at different parts of the hob, zoom them
in and out, and watch animations that present with high accuracy the activities to be
performed. The system can be used to train technicians or – for some actions – also
home users. Mistakes done when servicing a virtual hob are not followed by costs
unlike when using real devices for training.
The HMD used is Samsung Gear VR with the Galaxy S6 smartphone, which is now
a low-cost platform enabling presentation and interaction with reasonable quality. This
makes the platform available to small distributors of household equipment as well as
average users. The guide is a Unity3D application that implements the following
scenarios of repairing an induction hob. For every scenario, the guide presents inter-
active animated 3D models as well as movies.
(a) Checking the power connection (Fig. 6)
This scenario represents one of the most common failures. It includes an animation
of measuring power connection, by identifying appropriate terminals and demon-
strating how they should be connected.
Fig. 6. Checking the power connection of an induction hob.
(b) Checking coils and transistors

In this scenario, a technician verifies the state of coils and transistors (Fig. 7), which
are crucial for the correct work of the hob.
(c) Checking the overall work of the hob
This scenario allows a user to verify the entire process of switching on the hob and
cooking with it. The user is notified about possible problems that may occur and proper
solutions to the problems.
Fig. 7. Checking coils and transistors of the induction hob.
8 Conclusions and Future Works
The use of VR/AR gains increasing interest in manufacturing and marketing. These
technologies can be a key element and a direction towards innovation enabling man-
ufacturers to attract new customers. In this paper, we have presented five different
prototypes employing VR, AR and 3D web techniques for interactive 3D presentation
of products. Target visualization equipment ranges from smartphones, through HMDs
to large-scale VR installations. The presented prototypes have been developed in
collaboration with Amica S.A., which is the leading producer of household appliances
in Poland. These prototypes will be further developed and are expected to be deployed
in daily operation.
We plan to complete tests of users’ perception and interest in the developed
visualizations to assess the practical usefulness of the developed solutions. Further, we
plan to implement a method of implicit or explicit collection of users’ choices and
feedback to provide a tool for quick gathering of changing users’ preferences. Finally,
we plan to research creation of collaborative exhibitions and shopping spaces to enable
users to experience products in natural social setting (e.g., for families).
Acknowledgements. The authors would like to thank Amica S.A. for supporting this research
work and providing 3D models of products and graphics.
References
1. Griffin, T., Giberson, J., Lee, S.H., Guttentag, D., Kandaurova, M., Sergueeva, K., Dimanche,
F.: Virtual reality and implications for destination marketing. In: Travel and Tourism
Research Association International Conference, Quebec City, QC, Canada, June 2017
2. Klein, L.R.: Creating virtual product experiences: the role of telepresence. J. Interact. Mark.
17(1), 41–55 (2003)
3. Lee, K.C., Chung, N.: Empirical analysis of consumer reaction to the virtual reality shopping
mall. Comput. Hum. Behav. 24(1), 88–104 (2008)
4. Fransen, M.L., Verlegh, P.W., Kirmani, A., Smit, E.G.: A typology of consumer strategies
for resisting advertising, and a review of mechanisms for countering them. Int. J. Advert. 34
(1), 6–16 (2015)
5. Cappa, F., Del Sette, F., Hayes, D., Rosso, F.: How to deliver open sustainable innovation:
an integrated approach for a sustainable marketable product. Sustainability 8(12), 1341
(2016)
6. Walczak, K.: Structured design of interactive VR applications. In: Proceedings of the 13th
International Symposium on 3D Web Technology (Web3D 2008), pp. 105–113. ACM, New
York (2008)
7. Pellens, B., De Troyer, O., Kleinermann, F.: CoDePA: a conceptual design pattern approach
to model behavior for X3D worlds. In: Proceedings of the 13th International Symposium on
3D Web Technology (Web3D 2008), pp. 91–99. ACM, New York (2008)
8. Polys, N., Visamsetty, S., Battarechee, P., Tilevich, E.: Design patterns in componentized
scenegraphs. In: Proceedings of SEARIS. Shaker Verlag (2009)
9. Dachselt, R., Hinz, M., Meissner, K.: Contigra: an XML-based architecture for component-
oriented 3D applications. In: Proceedings of the 7th International Conference on 3D Web
Technology (Web3D 2002), pp. 155–163. ACM, New York (2002)
10. Visamsetty, S.S.S., Bhattacharjee, P., Polys, N.: Design patterns in X3D toolkits. In:
Proceedings of the 13th International Symposium on 3D Web Technology (Web3D 2008),
pp. 101–104. ACM, New York (2008)
11. Walczak, K.: Flex-VR: configurable 3D web applications. In: Proceedings of the Conference
on Human System Interactions (HSI 2008), pp. 135–140. IEEE (2008)
12. Alpcan, T., Bauckhage, C., Kotsovinos, E.: Towards 3D internet: why, what, and how? In:
International Conference on Cyberworlds (CW 2007), pp. 95–99. IEEE (2007)
13. Van Gool, L., Leibe, B., Muller, P., Vergauwen, M., Weise, T.: 3D challenges and a non-in-
depth overview of recent progress. In: Sixth International Conference on 3-D Digital
Imaging and Modeling (3DIM 2007), pp. 118–132 (2007)
14. Zahariadis, T., Daras, P., Laso-Ballesteros, I.: Towards future 3D media internet. NEM
Summit, Saint-Malo, France, 13–15 October 2008
15. Spagnuolo, M., Falcidieno, B.: 3D media and the semantic web. IEEE Intell. Syst. 24(2), 90–
96 (2009)
16. De Troyer, O., Bille, W., Romero, R., Stuer, P.: On generating virtual worlds from domain
ontologies. In: Proceedings of the 9th International Conference on Multi-Media Modeling,
Taipei, Taiwan, pp. 279–294 (2003)
17. Latoschik, M.E., Blach, R., Iao, F.: Semantic modelling for virtual worlds a novel paradigm
for realtime interactive systems? In: Proceedings of the ACM Symposium on Virtual Reality
Software and Technology (VRST 2008), pp. 17–20 (2008)
18. Chaudhuri, S., Kalogerakis, E., Giguere, S., Funkhouser, T.: Attribit: content creation with
semantic attributes. In: Proceedings of the 26th Annual ACM Symposium on User Interface
Software and Technology (UIST 2013), pp. 193–202. ACM (2013)
19. Walczak, K., Flotyński, J.: On-demand generation of 3D content based on semantic meta-
scenes. In: First International Conference on Augmented and Virtual Reality (AVR 2014).
LNCS, Lecce, Italy, pp. 313–332. Springer International Publishing (2014)
20. Walczak, K., Flotyński, J.: Semantic query-based generation of customized 3D scenes. In:
Proceedings of the 20th International Conference on 3D Web Technology (Web3D 2015),
pp. 123–131. ACM, New York (2015)
21. Flotyński, J., Walczak, K.: Ontology-based creation of 3D content in a service-oriented
environment. In: Proceedings of the 18th International Conference on Business Information
Systems (BIS 2015), Poznań, Poland, pp. 77–89. Springer International Publishing (2015)
22. Flotyński, J., Walczak, K.: Ontology-based representation and modelling of synthetic 3D
content: a state-of-the-art review. Comput. Graph. Forum 36(8), 329–353 (2017)
23. Asset Store: Unity Technologies (2017). https://www.assetstore.unity3d.com/en/
24. High Quality 3D Models: Scripts, Plugins and More! Highend3D (2017). https://www.
highend3d.com/
25. D Models for Professionals: TurboSquid Inc. (2017). https://www.turbosquid.com/
26. D ContentCentral: Dassault Systèmes (2017). https://www.3dcontentcentral.com/
27. Cortes, G., Marchand, E., Ardouinz, J., Lécuyer, A.: Increasing optical tracking workspace
of VR applications using controlled cameras. In: IEEE Symposium on 3D User Interfaces
(3DUI 2017), pp. 22–25. IEEE (2017)
28. Roup, M., Bosch-Sijtsema, P., Johansson, M.: Interactive navigation interface for virtual
reality using the human body. Comput. Environ. Urban Syst. 43, 42–50 (2014)
29. LaViola Jr., J.J.: Context aware 3D gesture recognition for games and virtual reality. In:
ACM SIGGRAPH 2015 Courses, pp. 10:1–10:61. ACM, New York (2015)
30. Piumsomboon, T., Lee, G., Lindeman, R.W., Billinghurst, M.: Exploring natural eye-gaze-
based interaction for immersive virtual reality. In: IEEE Symposium on 3D User Interfaces
(3DUI 2017), pp. 36–39. IEEE (2017)
31. Zielasko, D., Neha, N., Weyers, B., Kuhlen, T.W.: A reliable nonverbal vocal input
metaphor for clicking. In: IEEE Symposium on 3D User Interfaces (3DUI 2017), pp. 40–49.
IEEE (2017)
32. Shiratuddin, M.F., Zulkifli, A.N.: Virtual reality in manufacturing. In: Management
Education for the 21st Century, Ho Chi Minh City, Vietnam, 12–14 September 2001
33. Mottura, S., Viganò, G.P., Sacco, M.: Virtual reality for product layout configuration. In:
International Conference on Advanced Research in Virtual and Rapid Prototyping (VRAP
2003), Leiria, Portugal, October 2003
34. Hagge, L., Herz, A., Lang, S., Rupprecht, V., Sühl, S., Welle, N.: 3D visualization,
simulation and virtual reality in accelerator development. In: Proceedings of IPAC 2011, San
Sebastián, Spain (2011)
35. Barnes, S.: Understanding virtual reality in marketing: nature, implications and potential.
SSRN Electron. J. 1–50 (2016). https://doi.org/10.2139/ssrn.2909100
36. Reality – “How Reality Technology is Used in Marketing” (2017). https://www.realitytech
nologies.com/applications/marketing
37. X3D – Extensible 3D Graphics format – ISO/IEC IS 19775:2013. http://www.web3d.org/
standards
38. X3DOM – Fraunhofer-Gesellschaft. https://www.x3dom.org/
39. WebGL – OpenGL ES for the Web, The Khronos Group. https://www.khronos.org/webgl/
40. Vuforia – Augmented Reality for the Industrial Enterprise. https://www.vuforia.com/
The Effectiveness of Multimodal Sensory
Feedback on VR Users’ Behavior
in an L-Collision Problem
Sumin Kim1(&), Krzysztof Izdebski3, and Peter König1,2

1
Institute of Cognitive Science, Universität Osnabrück, Osnabrück, Germany
sumkim@uni-osnabrueck.de
2
Institut für Neurophysiologie und Pathophysiologie,
Universitätsklinikum Hamburg Eppendorf, Hamburg, Germany
3
SALT AND PEPPER Software GmbH & Co. KG, Osnabrück, Germany
Abstract. Virtual Reality (VR) is highly dependent on visual information,

although it offers multimodal channels for sensory feedback. In this study, we
compared the effectiveness of different sensory modalities in the context of
collision avoidance in the industrial manufacturing process. Participants per-
formed a pick-and-place task with L-shaped objects on a virtual workstation. In a
between-subject design each person performed one of four conditions: Baseline,
Auditory, Haptic, and Visual condition. We measured the timing and accuracy of
the performed actions. Statistical testing by an ANOVA showed a significant
main effect, i.e. a difference between the conditions. We observed the lowest
number of collisions in the auditory condition followed by the haptic, baseline
and visual conditions. Post hoc tests revealed a significant difference between the
auditory condition, the most accurate, and the visual condition, the least accurate.
This implies that giving additional feedback by the visual modality is not optimal
and utilizing a fully multimodal interface has increased effectivity.
Keywords: VR Multisensory feedback Collision Simulation
1 Introduction
Virtual Reality (VR) has evolved quickly in the last decade. The technical basis and
critical performance criteria have been largely improved and now allow for the
development of virtual environments with high immersion. Simultaneously, due to the
rising number of applications and enthusiasts, the price tag has dropped considerably.
This allows applications in new domains like entertainment, science, and industrial
applications. The industrial use of VR has its emphasis on simulating and prototyping
production processes in virtual environments. Notable features are the realistic ren-
dering of the environment including multimodal features, naturalistic behavior in the
VR by participants and dynamic feedback contingent on task performance. Thus, VR
tries to combine the best of both worlds and triggers a quantitative and qualitative
change in prototyping production processes. Boxplan, an example of industrial VR
software, is a virtual space where users plan their assembly stations at scale, create 3D
mock-ups and experience the assembly workflow. Thus, they can faithfully test the
https://doi.org/10.1007/978-3-030-18715-6_32
382 S. Kim et al.
layout concept in industrial and economic contexts. Importantly, feedback by the later
users is quickly incorporated leading to short turnaround times and a reduction in costs.
For such applications, users require more than just a realistic visualization, but the
direct feedback that directs the realistic physical behavior of the users and the inter-
action with virtual objects [1]. Therefore, the goal of such a VR-application can be
achieved only by realistic experience, as it is essential to recognize that the realism of a
training simulation influences training effectiveness.
Training in a virtual environment can take different forms. Performing complex
movements, such as in many sports disciplines, naturalistic feedback of desired move-
ment trajectories might be given. Other applications, where no single optimal behavior is
defined, might limit themselves to alarms when an error occurs. Specifically, a collision,
one of the most common physical interactions, can trigger such an alarm using multiple
modalities. In a natural environment, it might be seen, be audible, or be felt. However,
basic virtual environments confine themselves to the visual modality, i.e. when a col-
lision occurs the visible movement is stopped. This establishes a baseline condition.
Additionally, signals in other modalities, either in naturalistic form or as standardized
alarms may be supplied [2]. That is, the realistic sound of the collision could give
feedback on the erroneous movement. This, however, would require simulation of the
material properties, which is well beyond the scope of familiar virtual environments.
Therefore, standardized acoustic alarms are often used. Similar concerns apply to
feedback by the tactile modality [3]. To implement natural force feedback for free
movements is much more demanding than a simple vibration alarm. Still, modern
technology provides many choices for multimodal feedback. In this respect, multimodal
feedback has been investigated in different scenarios [4]. The effectiveness of tactile,
visual, and auditory warnings for the case of rear-end collision prevention in simulated
driving has already been demonstrated [5]. It found that the collision warning systems
consisting of different sensory feedback have a reliable effect on user’s behavior and
influence the number of collisions made [5]. A few studies already revealed that mul-
timodal feedback design could enhance motor learning and reduce workload, taking
advantage of each modality, which is especially beneficial for complex task and pro-
duction process in industries [2].
The importance of haptic feedback in VR is quickly growing (e.g., HaptX) and has
received considerable attention since the very early study of VR [6–8]. However, while
the comparison among general sensory feedback is competitively debated even in the
framework of VR, the effectiveness of their functions in different contexts has been
explored, but no general understanding is achieved [4]. Hence, the goal of our study is
to compare different sensory modalities regarding their effectiveness in collision
avoidance in VR.
2 Method
To analyze the effectiveness of different sensory-feedback modalities, we set up a study

that compares users’ behaviors in four different sensory-feedback conditions: a baseline
condition with naturalistic visual feedback, an auditory condition with an additional
auditory alarm, a haptic condition with an additional tactile alarm, and a visual
The Effectiveness of Multimodal Sensory Feedback on VR Users’ Behavior 383
condition with an additional color-changing visual alarm. We compare these conditions

in the L-collision problem. This problem describes a collision made by an L-shaped
object with other obstacles. When the head of the L-shaped object is behind the
obstacle, a behavior of pulling out the object causes collisions easily. Therefore it
provides a suitable environment to address the question of interest.
2.1 Participants
In our experiment, 65 volunteers (21 female, 44 male), aged between 19 and 35,
participated. All participants had an average or corrected-to-normal vision and did not
have any known neurological conditions. Due to a misunderstanding of the task
instructions, the data from two participants were excluded. In total, we measured 15
participants in the baseline condition, 15 participants in the auditory condition, 17
participants in the haptic condition and 16 participants in the visual condition. Each
participant was introduced to only one of the four conditions, and the condition was
randomly chosen before the participant was known to the experimenter.
2.2 Apparatus
For our study, we used a VR-ready PC with Nvidia 1070, Unity3D 5.6.3p2, NewtonVR
and HTC Vive HMD (110-degree field of view, 90 Hz, resolution 1080 1200 px per
eye). As we used the NewtonVR environment in our study, we decided to set a pure
NewtonVR condition with only its natural visual cue as our control condition (baseline
condition).
Fig. 1. (a) This figure describes the experimental setup scene of the shelf and the box.
A participant needed to pick up an L-shaped object, pull out of the shelf and place into the box to
complete a trial (top). (b) This is a picture of the first shelf L-shaped objects (5 per each story)
and obstacles (bottom). (c) Participants were given enough time to adapt to the VR environment
(right).
384 S. Kim et al.
2.3 Task
The task employed in this study was the L-collision problem (Fig. 1). A two-story shelf
with different-sized L-shaped objects was positioned in front of the user. Obstacles
were mainly consisting of two types: First, an obstacle with a minimal gap between the
ceiling of the shelf and itself so that the user could rotate the L-shaped object or pull it
to the side. Second, an obstacle that has a large enough gap between itself and the
ceiling, such that the user can apply any movements, for example merely lifting the L-
shaped object and pulling it out directly. The participants were instructed to pull out ten
L-shaped objects out of obstacles on the two-story shelf under one of the four different
feedback collision conditions. By putting the selected L-shaped object into the box
behind the user, the trial was considered to be completed. When users missed or
dropped the L-shaped object before they placed the object into the box, the trial was
recorded as a failure. Avoiding collisions with the given obstacles was not mentioned
so that the participants must recognize that by themselves via received feedback.
However, because the interacting object was L-shape, it was technically difficult to pull
it out and complete the task successfully when it stuck or collided with obstacles.
Therefore, by giving an instruction of pulling the L-shaped objects out of the obstacles,
the participants had to try to avoid the collision. The removal task was explained
individually to each participant. Participants was given one of the four different
feedback collision conditions: baseline, auditory, haptic and visual condition. In the
baseline condition, which served as the control condition, participants received no
feedback other than the natural visual cue. The visual cue here was given by default
physics setup of NewtonVR. The interacting object did not pass through the obstacles,
and participants could not complete the task without finding a way to avoid the
obstacles. All other conditions also included this type of natural NewtonVR visual cue.
On top of that, auditory, haptic and visual condition employed an additional modality
for feedback. Thus, these three conditions are multimodal feedback conditions
including the natural visual cue provided by NewtonVR setup as well. In the auditory
condition, an alarm sound played when the object touched any obstacles. In the haptic
condition, the controller performing the grabbing motion vibrated to indicate a colli-
sion. In the visual condition, participants received additional visual feedback via having
the color of the object being changed every time it touched another object, here we call
them obstacles, and reverting to its original color once it is no longer touching other
objects. The L-shaped object’s color material changed to black when it touched an
obstacle, and it changed back to its original color when it was moved away from the
obstacle.
2.4 Procedure and Analysis

We recorded the number and timing of collisions. Each collision was labeled with the
trial number and the index of the specific L-shaped object involved in the collision.
Completion of the trial was defined such that the L-shaped object interacted with the
collider of the box, which was placed behind the user. Successful completion was noted
as well as the number of failures to complete the task. In order to make the collisions
comparable in different conditions, consistent feedback types were provided throughout
each multimodal condition. For example, the same feedback color in visual condition,
the same beeping sound in auditory condition, and the same vibration frequency in
haptic condition were used. The analysis was performed using a one-way ANOVA and
Tukey Tests for post-hoc analysis. For outlier treatment, capping process was used by
replacing those observations outside the lower limit with the value of 5th percentile and
those that lie above the upper limit, with the value of 95th percentile before parametric
testing by an ANOVA.
3 Results
As a first step, we performed two controls that could potentially influence the inter-
pretation of our results. Specifically, we examined learning effects and compared the
level of difficulty for different objects. For this reason, two different sequences of the
number of collisions were visualized before the analysis of the number of collisions
among different sensory-feedback types.
Fig. 2. (a) The number of collisions for the different trials; (b) The number of collisions for the
different L-shaped objects; (c) Boxplot on the four different conditions to compare their mean
value occurred.
386 S. Kim et al.
As each participant performed the task in a pseudo-random sequence, we checked

whether a learning effect occurred over the trials. The Fig. 2a demonstrates that most of
the users made more collisions on their first trial than any other trial, and the least
number of collisions on their last trial. However, a higher index of the trial did not
always lead to fewer collisions. In the other trials, trial 2 to trial 9 seem to vary on the
number of collisions. On the analysis of the Fig. 2b, the fifth object (L5) seemed to be
the most challenging object for the users. It caused more than 20 collisions on average.
In comparison, the fourth object (L4) was the easiest, with less than five collisions on
average. However, besides these two extreme cases, the variation in the number of
collisions for different objects was moderate and, thus, added a limited amount of
variance to each task. Hence, we concluded that there was no strong learning effect that
occurred after the first trial in this task. Also, as all subjects handled all objects and
performed the identical number of trials the data might be well averaged over these two
variables.
For the next step, we focused on the differences between the conditions to explore
the effectiveness of multimodal feedback. The auditory condition resulted in the least
number of collisions, followed by the haptic condition, the baseline condition, and
finally the visual condition. For statistical analysis, a one-way between subjects
ANOVA was conducted to compare the number of collisions of the virtual L-shaped
objects and obstacles in the different conditions. There was a significant difference in
the number of collisions for the four conditions [F(3) = 2.9, p = 0.0424]. As we found
a statistically significant main effect of condition, we computed a Tukey post-hoc test.
Tukey HSD post-hoc test indicated that the mean score for the auditory condition
(M = 94.6, SD = 58.98) was significantly different than for the visual condition
(M = 151.03, SD = 88.32) [p < 0.05.]. No other significant pairwise differences were
found. Thus, we observe significantly different numbers of collisions as a function of
condition, and specifically that the number of collisions in the auditory feedback
condition is reduced in the pairwise comparison to the visual condition.
4 Discussion
With this experiment, we could demonstrate that the choice of modality influences the
effectiveness of multimodal feedback. Specifically, supplying the additional feedback
by another modality seems more effective than using the visual modality for natural
feedback and an alarm signal at the same time.
A prior study comparing the effectiveness between visual-auditory and visual-
tactile multimodal feedbacks on users in real-world task setup suggested that multi-
modal feedback is advantageous when compared to single modalities [4]. Specifically,
it showed that visual-auditory feedback is most effective when a single task is being
performed [4]. Another prior study also made a convincing case for the inclusion of
multimodal feedback for common direct manipulations such as the drag-and-drop and
showed that the inclusion of auditory feedback was common to conditions that
improved performance [9]. These results obtained in real-world setups match the
observations of the present study using a VR setup. Here, that the auditory feedback
condition, which includes the natural visual feedback, shows the best performance.
Furthermore, the visual feedback condition, which contained a visual alarm on top
of the natural visual feedback, performed significantly worse in our study. We spec-
ulate that the central focus induced by the task in VR reduced attention to the peripheral
vision and made the visual alarm less effective [10]. The visually alarming feedback
detection, therefore, might not function as effectively as it would do in non-constrained
real-world conditions. Another finding was that the controlled baseline condition with
the natural visual cue did not significantly differ from the visual feedback condition, if
anything it was slightly better. In other words, there was a large number of collisions
detected in the visual feedback condition, particularly compared to other conditions
including baseline condition. This provides evidence that a visual cue and additional
visual information convey a similar type of information. From this we can conclude
that both together do not improve the effectiveness of feedback.
These results are compatible with studies investigating the attentional bottleneck of
multiple modalities [11–13]. They report that in a dual-task setup the interference is
reduced when multiple modalities are involved. Also, further studies demonstrated that
multimodal feedback is advantageous when compared to single modalities in a variety
of task setups [4, 9, 14]. In this respect, natural visual cue and additionally designed
visual feedback of a color change are both mediated by the visual modality. This
explains why the visual condition, which is technically a single modality feedback
condition, performed worse than other multimodal feedback conditions.
A further reason for the increased effectiveness of auditory feedback might be that
participants are only adapted to realistic feedback. In the real world, when collisions
occur between two objects, auditory and haptic feedback are naturally generated, as
well as force feedback and the natural visual cue. However, the change of color of the
colliding object is rather artificial feedback. As our motor control is influenced by
internal representations of the actual and predicted states of our body and the external
world environment [15], such an artificial cue would not be predicted in the case of an
error, i.e., is harder to interpret. We speculate that the lack of natural internal repre-
sentations due to the unrealistic feedback, such as the changing of the color of the
object, can lead to ineffectiveness in task performance in VR.
In line with our study, there is a study that shows the utilization of the haptic
feedback in telepresence assembly task environment, whose setup is comparable to
virtual environments. Although it emphasized on the utilization of the haptic feedback,
it highlights that more realistic presence under haptic feedback was achieved by other
modalities such as a visual bar graph or an auditory stimulus [8], supporting the
effectiveness of the auditory feedback.
Another finding is in accordance with our result, showing the efficacy of multi-
modal feedback in general [4, 9, 14]. This was reached by a few studies, comparing the
effect of different modalities on user’s performance. One study specifically found out
that the multimodal combination of visual-auditory feedback yields more favorable
results regarding performance than visual feedback alone in single task scenarios under
normal workload conditions [4]. Also, our study expanded the result of a joint task
study in a non-VR condition [13] to a VR condition, showing that auditory displays are
a viable option to receive task-related information in virtual reality as well.
To that end, our study demonstrates that different types of feedback should be
considered depending on the different contexts of VR applications in order to optimize
388 S. Kim et al.
their effectiveness. Notably, our results cast a new light on the function of sensory
modalities other than vision in VR. However, as pointed out in other similar studies,
with varying workloads, different modalities could offer additional advantages. Hence,
our research suggests further study into investigating these findings in more specific
contexts or different tasks in order to apply them to particular practical cases.
Acknowledgments. We gratefully acknowledge the support by the project ErgoVR (BMBF,

KMU Innovativ V5KMU17/221) and the SALT AND PEPPER Software GbmH & Co.KG.
References
1. Ragan, E.D., Bowman, D.A., Kopper, R., Stinson, C., Scerbo, S., McMahan, R.P.: Effects of
field of view and visual complexity on virtual reality training effectiveness for a visual
scanning task. IEEE Trans. Vis. Comput. Graph. 21(7), 794–807 (2015)
2. Sigrist, R., Rauter, G., Riener, R., Wolf, P.: Augmented visual, auditory, haptic, and
multimodal feedback in motor learning: a review. Psychon. Bull. Rev. 20(1), 21–53 (2013)
3. Hayward, V., Astley, O.R., Cruz-Hernandez, M., Grant, D., Robles-De-La-Torre, G.: Haptic
interfaces and devices. Sens. Rev. 24(1), 16–29 (2014)
4. Burke, J.L., Prewett, M.S., Gray, A.A., Yang, L., Stilson, F.R., Coovert, M.D., Elliot, L.R.,
Redden, E.: Comparing the effects of visual-auditory and visual-tactile feedback on user
performance: a meta-analysis. In: Proceedings of the 8th International Conference on
Multimodal Interfaces, pp. 108–117. ACM (2006)
5. Scott, J.J., Gray, R.: A comparison of tactile, visual, and auditory warnings for rear-end
collision prevention in simulated driving. Hum. Factors 50(2), 264–275 (2008)
6. Burdea, G.C.: Keynote address: haptics feedback for virtual reality. In: Proceedings of
International Workshop on Virtual Prototyping, Laval, France, pp. 87–96 (1999)
7. Srinivasan, M.A., Basdogan, C.: Haptics in virtual environments: taxonomy, research status,
and challenges. Comput. Graph. 21(4), 393–404 (1997)
8. Petzold, B., Zaeh, M.F., Faerber, B., Deml, B., Egermeier, H., Schilp, J., Clarke, S.: A study
on visual, auditory, and haptic feedback for assembly tasks. Presence: Teleoper. Virtual
Environ. 13(1), 16–21 (2004)
9. Jacko, J.A., Scott, I.U., Sainfort, F., Barnard, L., Edwards, P.J., Emery, V.K., Kongnakorn,
T., Moloney, K.P., Zorich, B.S.: Older adults and visual impairment: what do exposure times
and accuracy tell us about performance gains associated with multimodal feedback? In:
Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, pp. 33–
40. ACM (2003)
10. Khan, A.Z., Blohm, G., McPeek, R.M., Lefevre, P.: Differential influence of attention on
gaze and head movements. J. Neurophysiol. 101(1), 198–206 (2009)
11. Wahn, B., König, P.: Can limitations of visuospatial attention be circumvented? A review.
Front. Psychol. 8, 1896 (2017)
12. Wahn, B., König, P.: Is attentional resource allocation across sensory modalities task-
dependent? Adv. Cogn. Psychol. 13(1), 83 (2017)
13. Wahn, B., Schwandt, J., Krüger, M., Crafa, D., Nunnendorf, V., König, P.: Multisensory
teamwork: using a tactile or an auditory display to exchange gaze information improves
performance in joint visual search. Ergonomics 59(6), 781–795 (2016)
14. Lee, J.H., Spence, C.: Assessing the benefits of multimodal feedback on dual-task
performance under demanding conditions. In: Proceedings of the 22nd British HCI Group
Annual Conference on People and Computers: Culture, Creativity, Interaction-Volume 1.
British Computer Society, pp. 185–192 (2008)
15. Frith, C.D., Blakemore, S.J., Wolpert, D.M.: Abnormalities in the awareness and control of
action. Philos. Trans. R. Soc. London. Ser. B 355(1404), 1771–1788 (2000). https://doi.org/
10.1080/00140139.2015.1099742
Intelligent Systems in the Railway Freight
Management
Michal Balog1, Hanna Sokhatska2, and Angelina Iakovets1(&)

1
Faculty of Manufacturing Technologies with a Seat in Presov,
Technical University of Kosice, Bayerova, 1, 08001 Presov, Slovak Republic
angelina.iakovets@tuke.sk
2
Faculty of Entrepreneurship and Law, Kyiv National University of Technology
and Design, Nemirovich-Danchenko, 2, Kiev 01011, Ukraine
Abstract. The article is devoted to the problems of improving the management

system of railway freight transportation based on the concept of Industry 4.0.
During the study, was identified the main competitor of rail traffic - automotive
freight transportation. Investigation of the rail freight market has shown that
there are a number of problems that prevent them from taking a leading position.
The main part of research was directed on the European market, especially on
the Slovak Republic. Was considered modern monitoring systems of transport
logistics and was determined that the most advanced and responsive to the
demands of the modern consumer is tracking technology based on GPS -
navigation control technology. Proposals were developed to improve the effi-
ciency of freight wagons monitoring system by using electronic control unit
based on geolocation technology integrated into carriage, they will signalize
about condition of the freight in active and passive form. Such sensor will be
able to integrate to the IoT system through a specialized Internet platform that
also supports mobile version program. Were justified the economic feasibility of
introducing this system to European rail freight and development prospects of
the railway transportation in the context of the Industry 4.0.
Keywords: IoT Industry 4.0 Freight rail transportation

Freight control system GPS monitoring Transport logistics
1 Introduction
Obviously, that era of Industry 4.0 defines a radically different management format
based on high technology. The key elements of such control system are cloud com-
puting, combining cyber physical systems and the Internet of Things space in real time
to provide communication when creating values in the format of “smart” production.
Consumer Internet of Things quickly gained a position in the technology market,
having firmly entered the life of the average person in the form of simple and con-
venient solutions that provide many advantages and a new quality of life. According to
forecasts of consulting Juniper Research [1], by the end of 2018, the number of Internet
of Things devices in the world will increase to 21 billion, and by 2022 it will exceed 50
billion. This forecast is confirmed by the McKinsey Global Institute research, which

https://doi.org/10.1007/978-3-030-18715-6_33
Intelligent Systems in the Railway Freight Management 391
says that world economy due to the introduction of the Internet of things by 2025 can
be from 4 to 11 trillion US dollars, and a significant part of this amount will be
provided by the introduction of IoT in production [2, 3]. It is obvious that companies
that will be able to maximize the opportunities of the digital revolution will occupy a
leading position in the market. The elements of the industrial Internet of things are also
widely used in the freight industry. So, one of the priority directions is monitoring of
movement of goods in time and space. For this purpose, various methods of contactless
traffic control, GPS navigation systems, and RFID technologies are used, which helps
to create full volume of information that costumer needs.
Considering rail freight, as an industry, in terms of flexibility to the needs of the
consumer, still stay behind competitors in this economic sphere. The customer expects
that information about transporting freight will be always on hand it means that
information will be available not only on website, but also in mobile application. Not
every company in European area able to give such a volume of information as com-
petitors in the market. The driver of the locomotive have no information about the
cargo during transportation, so all of the problems that have arisen through the route
will be detected only at the points where the freight will be unloaded from the wagons.
Real state of updating information about transportation also is very low, it caused by
the fact that the information is entered into the software only upon the arrival of the
locomotive at the cargo transfer station.
All this reasons are contribute today’s market position of the railway freight. Since
rail transportation today is not a market leader, even despite their advantages over
competitors, this becomes a reason for researching enterprises in this industry and
developing proposals for improving their competitive position.
2 Analysis of the Freight Market in Europe
2.1 The Modern Level of Rail Freight Development

The railway freight market is characterized by high conservatism, which significantly
reduces the consumer attractiveness of this segment from the point of view of safety
and transparency of the transportation process.
Rail transport takes important role and has a considerable potential for development
in the field of freight. According to the EU Agency for Railways, the length of railways
is 236 thousand km (Fig. 1), and the volume of cargo transported according to esti-
mates by the European Statistics Committee in 2016 was more than 430 million tons
(Fig. 2).
The European railway system is so well developed (Fig. 1) which allows trans-
porting large volumes of goods between major cities in the cheapest and safest way.
392 M. Balog et al.
Fig. 1. Rail Freight Corridors in Europe – map 2018 [4]
Fig. 2. Evolution of EU-28 rail freight transport for main undertakings, 2006–2016 [5]
The carriage of goods by rail has a number of undeniable advantages, such as

efficiency; high loading capacity of wagons; the ability to transport light and bulky
cargo (more than 120 m3), a variety of rolling stock. However, in recent years, as can
be seen (Fig. 2), a decrease in the loading of marketable rolling stock in the region as a
whole is observed.
Fig. 3. Transport of goods by freight transport [6]
The main disadvantages are: very low transport accessibility for senders and
recipients; real-time updating information about the route; the impossibility of online
monitoring of the technical condition of all parts of the train.
Some disadvantages are not so obvious, for example, the interaction of a loco-
motive and wagons, what causes technological difficulties. As an example of the sit-
uation, the data are illustrated by the schedule of cargo volumes transportation on the
territory of the Slovak Republic, by type of transport (Fig. 3).
The presented graph illustrate that the volume of road transport is 3.5 times higher
than volume of the rail transport, which is confirmed by European statistics. These
tendencies are also considered in other literary sources [7, 8]. It is also important to
note the fact that in the long run, rail transport is much less subject to fluctuations under
the influence of such market factors as the price of fuel (Fig. 3). This is a weighty
argument in favor of the reliability of rail transport over other types of transportation.
A comparative analysis of the characteristics of the transportation of goods is presented
in Table 1.
According comparison from the table, road convey has a number of significant
advantages such as the possibility of monitoring vehicle location and real-time com-
munication with the driver, real-time monitoring of technical condition of vehicle, this
number of advantages of road transport can be a reserve for development in the field of
rail transport. Railway transport can improve its position on the market by upgrading
communication systems, also by installation of self-diagnosis systems to drivers cab
and wagons, at the same time with a help of providing mobile app real-time monitoring
program.
In the railway industry, today, the appropriate types of wagons are used to transport
goods (Table 2).
From the table below we can conclude that the range of railway transport rolling
stock is characterized by a wide variety of wagons and allows satisfying the needs of
any customer.
394 M. Balog et al.
Table 1. Comparative advantages and features of cargo transportation by road and railway
[formed on 9]
Type of Advantages Disadvantages
transportation
Road Fast delivery Transport subjects to traffic delays
Cost effective Transport subjects to breakdown
Ideal for short distances Goods susceptible to damage
Ideal for transporting through careless driving
perishables Bad weather
Easy to monitor locations Driving regulations can cause delays
Easy to communicate with
driver
Ideal for sending by courier
Rail Fast delivery Subject to unforeseen delays
Capacity Reliance on operator’s timetable
Cost effective Suppliers/customers are not
Safe mode of transport always located near a rail freight
Reliable depot and delivery can be costly and time
Eco-friendly transport consuming
Table 2. Compliance of cargo types to railway wagons [10]

Cargo type Wagon
Packaged goods with manual and Covered wagons, containers
mechanized loading
Heavy loads, oversized loads Platforms and conveyors
with crane loading
Bulk cargoes Using specialized rolling stock
Wagons and other specialized Highly specialized wagons
goods
Cargo requiring special Thermoses and isothermal wagons as part of refrigerated
temperature conditions sections (depending on the type of cargo)
In general, an analysis of the industry’s capabilities suggests that rail freight ser-
vices in Europe have significant competitive potential but they need to be developed
and modernized as one of the most promising types of transport.
2.2 Control and Navigation Systems in Railway Logistics

The success of the development of any system today is determined by the level of
technologies used, flexibility and ability to adapt to a changing external environment.
This necessitates the search for and implementation of the most innovative techno-
logical solutions in the context of the concept Industry 4.0 to ensure safety and control
traffic. Exploring the current state of the freight management system on Europe’s
railways, it can be noted that there is considerable potential in introducing intelligent

navigation and control systems using the Internet of things.
In particular, it is reason to study the existing rolling stock management system
from the point of view of “smart” control in real time. Turning to the current state of
controlling and communication systems of railway freight and according to industry
publications [4, 11, 12] - modern traction and cargo railway transport in Europe is
equipped by control system with GPS-R control and other types of communication
with locomotive. This technology provides support for safe movement and coordina-
tion of train movement within the framework of the European area with a help of ETCS
(European Train Control System), as well as the European Rail Traffic Management
System (ERTMS).
The system determines the location of the train, the distance between trains, con-
trols the maximum allowed speed at each section, calculates the braking distance
(depended on speed and traveled distance), and compares the route data with the
technical speed abilities of the train. Reception, processing and transmission of all
necessary information are carried out by a complex of floor and on-board devices. The
action of the system is aimed at safely reducing and passing the interval. At the same
time, the main instruments of this system are located directly in the driver’s cab.
The scientists were trying to improve the tracking process of rolling stock,
researchers of the European project MODTRAIN proposed to upgrade the control
panel in the cab of a high-speed train driver for passenger traffic - MODLINK/EUCAB
Drivers desk layout [13]. According to the project, the driver’s cab should be equipped
with a control display for monitoring train status in real time, a diagnostic display for
displaying possible deviations and other modules for timely control of the transport
operation. With a number of advantages, this development is currently still in the
approbation stage, and moreover, it is not adapted to the conditions of freight loco-
motives, which makes it impossible to directly integrate it into the transport
infrastructure.
It should be noted that many of the innovative proposals to date have not yet found
ways to be implemented in the EU rolling stock management system. The issue of
modernization of the transport system of the Slovak Republic is especially acute. So,
according to the analysis conducted by Ministry of Transport and Construction of the
Slovak Republic, were identified main problematic spheres of this market. The dom-
inant position in the Slovak railway market belongs to ZSSK CARGO, that’s why the
most of statistical data will be about this organization.
It was determined that the freight rail transport is not a sufficiently competitive
range of services offered (compared to road transport) because of lower flexibility in the
provision of services and pricing for the services provided; longer delivery times
compared to road transport; not enough flexible for external changes (regarding
ZSSK CARGO and its customers); unfinished electrification of the transport process,
relatively outdated fleet of wagons and its unsatisfactory structure (absence of timely
control of the technical condition of carriages); repair costs are not reduced. Also
problematic areas of activity are: controlling process of the carriages technical condi-
tion; mileage control; monitoring of cargo in the container; the lack of communication
locomotive and wagon; lack of control of movement of each carriage; problems with
396 M. Balog et al.
communication during bad weather; limited customer access to cargo information. All
these items cause low demand for services of this type of transportation [14].
In support of these provisions, one can give an example of the process of moni-
toring the movement of goods on the website of ZSSK CARGO, the leader of the
Slovak railway traffic market [15]. The company provides information about the
location and movement of goods of the customer according to the contract in the form
of a table (Fig. 4).
Fig. 4. Transport timetable ZSSK CARGO [15]
This figure shows information about check in of locomotive at stops along the route
where the cargo is being transferred. The data of this table are not formed on the basis
on GPS-fixing of the cargo location, but only after the locomotive arrives to the transfer
station; all this evidence not fully using of train navigation system. Data about the
movement along the route are available only when locomotive of freight transfer will
pass the checkpoint, it means: in the same time when dispatcher will fill the table. The
customer can see via information only on Internet, according to the shipping list.
Should be noted this type of provision of information on the composition is sig-
nificantly inferior in terms of information content to that of road freight carriers. It is
obvious that full access to information about cargoes will provide significant com-
petitive advantages to haulers, making this type of transportation more attractive,
ensuring high reliability and transparency of the transportation process.
2.3 Directions of Modernization of the Railway Freight Management

System
To increase the demand for rail transportation among European customers, it is nec-
essary to raise the level of attractiveness of the service, ensuring high reliability of
transportation while maintaining an attractive pricing policy in the freight market.
One of the priority directions of development of this industry is the introduction of
a rolling stock management system based on the smart technologies of the industrial
Internet of things, when separate state monitoring and diagnostics devices are com-
bined into a single system to ensure the new quality of process management based on
the principles of Industry 4.0. The experience of introducing such systems in
production showed a number of advantages of a strategic order, and the practice of

using elements of smart management systems in road haulage suggests the possibility
of using this approach to form a new generation of transport systems based on the
Internet of things.
To date, the most important priority in the management of cargo transport on
railways is to ensure efficient communications between the main locomotive and freight
wagons during the route. This task does not yet have a practical solution, which is a
significant disadvantage for the carrier and the customer, since it does not allow
controlling the condition of the cargo during transportation.
Some attempts to improve the situation are already being made on the market.
Thus, according to the online publication Railway Gazette [16], Frequents plans to
deploy a national GSM-R network to support communication between dispatch ter-
minals and nodes using the Nokia IT platform for management software.
In order to prepare the studied railway infrastructure for possible future innova-
tions, it is necessary to establish a system of diagnostic and communication between
the car and the locomotive as well as with the control center. In this article proposes
installation of an electronic control unit into the carriage and locomotive, to collect
information about self-diagnosis as well as to improve communication and the data
collection. This control unit will create brand new types of wagon based on smart
technologies, which in the future help to create m2m real-time communication in the
context of IoT.
3 Improvement of the Management System in Railway Cargo

Transportation Based on the Implementation of IoT
Technology
3.1 Experience in the Implementation of Navigation Systems
in Transport Logistics
Transport logistics, in modern times, is impossible without controlling the movement
of vehicles, planning their routes, forming delivery schedules and other related activ-
ities. Considering the trucking, we can note the main types of programs that perform
these functions: PROLOGISTA, 1C: Vehicle Management, 4logist, TransTrade, Trans-
Manager, ABM Rinkai TMS, BasketCargo, Expedit, Cargo.24, and etc. The railway
freight transportation firms use such programs as: GVD, SGVD, ŽDC, ZONA, MET,
EDD, PMD, PIS and another. The basis of all these programs is the evidence of the
movement of vehicles and its stock with the help of active and passive geolocation
systems, for further planning of the main activities of freight transportation companies.
It should be noted that the above-mentioned programs of trucking companies, work
with large amount of information than rail freight companies, and are associated with
the work with a number of scanners. Trucking database formed not only on the
computers of the company but also on mobile applications, which is more effective and
intelligent.
The issues of improving navigation and traffic control systems, of course, are
always in the focus of the stakeholders of the freight process, since they determine the
398 M. Balog et al.
level of quality and key characteristics of the service. The degree of development of the
transport infrastructure and the quality of transportation directly determines the com-
petitiveness of companies - carriers and capital inflows into the development of the
industry as a whole.
Transport enterprises are actively introducing specialized software to control the
movement of vehicles, planning routes, coordinating the process of cargo transporta-
tion. Many carriers try to make the service more transparent by providingcustomer-
swiththeabilitytomonitorallkeyparametersoftransportation in real time. For this
purpose, software and cloud resources are actively used, which provide monitoring of
the movement of an object from any computer or mobile device, which is of particular
importance for freight haulage companies and organizations engaged in passenger
transportation. Vehicle monitoring is based on the use of telemetric equipment [17, 18].
Usually it is a small device that looks like a USB flash drive that integrates into the
car’s OBD-connector.
The principle of operation of transport monitoring and control systems consists
mainly in tracking both spatial and temporal coordinates of the vehicle. There are two
ways of monitoring activity: on-line data transmission and off-line data transfer (in-
formation is read from the tracking device and analyzed after the vehicle arrives at the
control room).
This control system allows the carrier and the customer to track the movement of
goods, making this type of transportation more manageable in real time. It is no
coincidence that the most attractive for transportation industry is development and
implementation of innovations in cargo management, which gives indisputable com-
petitive advantages to key players in the transportation market. To implement the best
solution, let us consider in more detail the principles of the work of navigation systems
in the automotive industry.
The management of transportation processes is a key element in the organization of
any logistics system. First electronic systems for monitoring the condition of the car
were introduced by Society of Automobile Engineers USA. Such control systems
began to include GSM/GPRS/GPS monitoring systems, which made it possible to
diagnose not only the technical parameters, but the location of the monitored object.
Also, there are GPS modems which use the NMEA, TAIP, Enfora binary type
exchange protocols. The modem exchanges NMEA messages in GGA, GLL, GSV,
GSA, RMC, VTG formats with a central server. It also has the function of storing GPS
messages in the nonvolatile memory of the module. The modem can work with ten
servers at the same time. In GPS mode, the following functions are also supported: a
virtual odometer, fixed geophones, and a format of buffered messages [19]. Such types
of modems are integrated into the vehicle control system and have the ability to inform
the driver through the onboard computer. Plurality of another tracking systems for
vehicles are used to determine the current location of the vehicle with an accuracy of
meters, but also to view the history of its movement, as well as perform a number of
other functions. Thus, modern tracking devices enable voice communication with the
driver, can monitor the environment, and receive notifications about the consumption
of gasoline or the deviation of the vehicle from the specified route.
An alternative GLONASS tracking system has been developed in the Russian

Federation. Their capabilities are used in trackers, have significant similarities, but are
still not identical. The key differences are as follows:
– Methods of using radio frequencies: in GLONASS - a more secure, but also more
resource-intensive channel FDMA (“dedicated line” principle), in GPS - a less
protected and stable CDMA (coded multiple access), but more economical.
– The principle of the location of satellites: the Russian satellite constellation is
independent of the rotation of the Earth; American - requires continuous synchro-
nization and adjustment of the satellite orbit.
– Accuracy in different latitudes: due to the higher orbit of Russian satellites, which
provide greater than GPS, accuracy in north latitudes, in particular, in the area of
Scandinavian countries.
All of the above means that a modern system of satellite transport monitoring can
use both navigation systems. Trackers that work with both GLONASS and GPS-
modules can improve the overall accuracy of object detection (up to 1.5–3 m) and
achieve stable and high-quality work even in difficult conditions of modern cities with
dense high buildings, which are strongly complicates the passage of signals from
navigation satellites [20]. Many of the currently proposed GPS trackers and controllers
have an open protocol for interacting with the server. This makes it possible to con-
figure operating modes using SMS, CSD or GPRS connections [17]. All technological
solutions offered on the market today that are responsible for monitoring and con-
trolling the movement of vehicles have one goal - to quickly provide reliable infor-
mation about the car and its location, which can then be applied in accordance with the
client’s business needs.
3.2 Opportunities to Upgrade the Management System of Railway

Transportation Based on Smart Technologies
To increase the attractiveness of railway transportation in the freight transportation
market, it is necessary to bring the freight management system in line with the prin-
ciples of “smart” management, ensuring continuous monitoring and the ability to adjust
system parameters in real-time. Such an approach is possible with the active integration
of IoT technologies into in all activities of such market. The need to introduce inno-
vations in this area is also evidenced by the analysis of the railway accident rate [7]. As
noted in this study, the main problem of analyzing incidents on railways is to obtain
information about the damage at the time of the accident. Another important part of
monitoring the status of the rolling stock is diagnosis of the braking system of the car
and suspension. Unlike road transport, which necessarily passes the technical control of
the state at certain intervals of run, the wagons do not have a similar tracking mech-
anism. For this reason, the technical condition of the suspension is still determined by
employees of railway stations with the help of a sound test (hammer control) of car
suspension. The photo (Fig. 5) shows wagons used on European territory for the
transport of goods.
400 M. Balog et al.
Fig. 5. Wheel wagon system
As photo shows, the used wagons do not have control sensors, as well as instru-
ments indicating the state of the braking system or mileage, etc. The absence of such
devices makes the rolling stock more vulnerable to unforeseen disruptions and acci-
dents, which significantly reduces the level of transport safety. According to the
analysis conducted by Ministry of Transport and Construction of the Slovak Republic
was proposed to improve the monitoring system by installation the electronic unit to
each wagon to record information about the state of the wagon and to transmit signals
about the presence of it. The best place for proposed electronic unit, which will collect
data, will be designated place on wagons scheme bellow (Fig. 6).
Fig. 6. Sensor control system
Regardless of the connection method, the most vulnerable part from the point of
view of the design of connecting parts of the car is the lower part of the body, because
during movement, the cars often collide when passing difficult sections of the road,
therefore the sensor in the specified place will respond better to changes along the
route. The research of market offers showed that there are several solutions from
different companies from different countries. One of such devices is offered by the
Bosch Company, their device will have an active GPS system and an alarm about
unauthorized opening of the container. Bosch also proposes to use their transportation
planning program, where the possibility of planning repair works will be realized with
the help of mileage data taken from the sensor. In another hand, all ideas of Bosh are
not realized in real, that’s why were decided to take real GPS control units. Due to this
was reviewed Schturman Monitoring Company’s device. This electronic control unit
includes: GPS and GLONASS navigation system, long life battery, FLS (fuel level
sensor) - upon condition of installation on the locomotive, container opening alarm
sensors. On another side there is device from Intelli Company’s, which has all men-
tioned functions and more:
– Wireless internet standard GSM/GPRS (850/900/1800/1900 MHz) (allows real
time tracking on computer and mobile phone or tablet).
– IP-standard protective housing.
– SMS tracking.
– Setting the position update frequency.
– Under condition of changing of direction - position update.
– Getting the location as an address.
– Check vehicle status.
– Audio monitoring (built-in microphone allows to hear what is happening around the
device).
– Built-in memory for offline recording data (route and another).
– Control of additional parameters.
– Sleep mode (for power saving).
– Getting started on schedule.
– Installation of virtual control zones (when wagon will exit from route zone - will be
sent an alert).
– Alarming system, what signalized about starting the movement or appeared acci-
dent, or speeding.
– Optimized data transfer protocols. Reduced consumption of GPRS traffic.
– Set of trusted numbers to which alarm messages will be sent.
– Intellectual control of the trajectory of movement (if it located in drivers cab).
– Collection the data to the memory buffer.
– Remote motor lock and/or control of an additional relay (only in locomotive).
– Battery low alarm [34].
Fig. 7. Types of wagon connections

402 M. Balog et al.
Such sensors are proposed to be mounted to monitor the condition of the loco-
motive and wagons and to create information communication net. There are several
main ways of connecting wagons, depending on their types (Fig. 7).
Types of wagon connections: (a) traditional connection by electric cable [21];
(b) chain connection with hydraulics, without electricity [22]. Due to this, arise two
abilities to implementations of propose: connecting device by electrical cables and
without them. Electricity will help to support continuous work of mounted GPS
electronic control unit. Each sensor will be connected with control block, which, using
satellite communications online, will be able to transmit information about the status of
wagons not only to the train driver, but also to dispatchers, as well as to freight
customers. The assumed scheme is depicted below (Fig. 8).
Smart
wagon
Fig. 8. GPS navigation system of railway transport using tracking sensors
Introduced GPS-modular system will allows monitoring the location of freight in

real time, as well as the mileage of the carriage, thus planning its maintenance. Pro-
posed diagnostic system create smart carriages park, where the smart technology
caused by self-diagnostic GPS control unit.
Taking into account the fact that the tracker has built-in memory, it is proposed to
write down the main accompanying document, which is drawn up when the goods are
sent and gives full information about the sender and details of the goods being sent. For
more extensive information about real condition of train and it’s freight, it is proposed
to connect the special tracker with long autonomous operation, FLS (fuel level sensor),
container opening sensors, SOS button, shock sensor and engine blocking relay - it
means to buy all components from proposed package of Intelli Company. Subject to
the installation of a full set of equipment will open brand new abilities to manage ZSS
Cargo. Will allowed opportunities for creating mobile application, in which will be able
online monitoring current state and cargo movement with data of the shipping
document.
4 Prospects and Features of the Introduction of a Smart

Monitoring System of Railway Freight
According to preliminary estimates, the implementation of the proposed wagon

movement monitoring system fully pays for itself for a period of up to 5 months,
according to the calculations of the proposed supplier of devices. Thus, the introduction
of sensors can significantly save on fuel and vehicle maintenance. By monitoring the
mileage of vehicles, fuel consumption will reduce, and the service life of equipment
will increase. For this reason, the efficiency of vehicles will increase and, as a result,
will help to prevent unforeseen repair costs and will provide an opportunity to conduct
a more extensive analysis of the enterprise.
The exact period of return on investment in the system depends on many factors,
and for each specific case an individual approach is necessary when calculating it.
Based on the analysis of the costs of the Ministry of Transport and Construction of the
Slovak Republic, it can be said that such equipment will help to take a leading position
for ZSSK CARGO in the freight market and removed all the shortcomings discovered
before also significantly reduce most of the costs and would create smart management
system.
The implementation of the proposed traffic monitoring system will also enable
receiving the information about the current location of trains and the status of wagons
and cargo, including information about temperature, as well as the communication of
the locomotive with wagons and dispatch departments, as well as properly setting the
parameters of the device, it will be possible to have all the detailed information in the
mobile application online and offline mode. As a result, the organization of trans-
portation will become more efficient, logistics processes will be optimized, and the cost
of transportation will decrease. At the same time, the function of preliminary assess-
ment of conditions will remind the operating service about tool deterioration, what will
allow to plan repair in advance. Moreover, GPS and burglar alarm functions ensure
additional security for the goods being transported. Would be possible to reduce
downtime and would help firm to become more flexible for customers needs and to
conditions of Industry 4.0. Modern management technologies will allow to create IoT
system of communication. The main advantage of the proposed devices is that the
company does not need to drastically change the existing equipment, but only to carry
out installation work and program the devices for those functions that are needed for a
particular enterprise. It can be speed up by practical implementation of the proposed
monitoring system; it is possible to adapt the existing Wagons-Control system [35].
Provided that the company is not able to independently implement such technologies -
the manufacturer of these devices offers its assistance at a require mental price and,
more importantly, at a lower price than alternative American and European firms.
404 M. Balog et al.
5 Conclusions
The main advantage of the proposed devices is that the company does not need to
drastically change the existing equipment, but only to carry out installation work and
program the devices for those functions that are needed for a particular enterprise. It
can be speed up by practical implementation of the proposed monitoring system; it is
possible to adapt the existing Wagons-Control system [35]. Implementation of pro-
posed device will create smart self diagnostic vehicle what fill the main ideas of
modern industry. If the railway cargo company is not able to independently implement
such technologies - the manufacturer of these devices offers its assistance at require
mental price and, more importantly, at a lower price than alternative American and
European firms.
The implementation of all the proposed solutions will allow railway enterprises to
take a qualitatively new level in the industry and also significantly approach to present
period of Industry 4.0 and open up the following prospects for future upgrades. So the
synergistic effect of all changes will lead to an improvement of rail traffic in general.
Acknowledgments. Research reported in this paper was supported by EU Structural Funds

within the project ITMS code 26220220160. “Support and dissemination of the Research Center
for Transport “CVD-PLUS”.
References
1. Juniper Research. https://www.juniperresearch.com/
2. McKinsey Global Institute. https://www.mckinsey.com/featured-insights/digital-disruption
3. Da Xu, L., He, W., Li, S.: Internet of Things in industry: an overview of key technologies
and trends. In: Control Engineering Russia, IIoT, April 2017. https://ptelectronics.ru/stati/
4. EU Agency for Railways – RNE. http://www.rne.eu/rail-freight-corridors/rail-freight-corrid
ors-general-information/
5. Railway freight transport statistics. Eurostat. https://ec.europa.eu/eurostat/statistics-explain
ed/index.php/Railway_freight_transport_statistics
6. Statistical office of the Slovak Republic. https://slovak.statistics.sk/wps/portal/ext/home/!ut/
p/z1
7. Rosová, A., Balog, M., Šimeková, Ž.: The use of the RFID in rail freight transport in the
world as one of the new technologies of identification and communication. Acta Montanist.
Slovaca 18(1), 26–32 (2013). ISSN 1335-1788
8. van Leijen, M.: Global rail freight volumes expected to rise in next two years. Railfreight.com
online magazine for rail freight professionals, 16 Aug 2018. https://www.railfreight.com/
business/2018/08/16/global-rail-freight-volumes-expected-to-rise-in-next-two-years
9. Serroukh, Y.: Modes of transport in logistics. In: Business English Logistics & Transportation
Engineering Seminar (2011)
10. Railway transportation to the CIS-States, Rail freight, GTM Solution. http://www.gtm-
solution.com
11. European Train Control System. https://en.wikipedia.org/wiki/European_Train_Control_
System
12. European rail traffic management system. EU Agency for Railways. https://www.era.europa.
eu/activities/european-rail-traffic-management-system-ertms_en
13. Steinicke, W.H., Meissner, T., Schober, M., Gelbert, L.: Future European-wide harmonized
train driver’s cab EUCAB – an essential result of the integrated project modular train.
Researchgate (2006)
14. A unified vision of the rail sector in SR, MDVRR SR. Vlaky.net (2018). https://www.vlaky.
net/zeleznice/spravy/
15. Customer portal ISP, Tracking of wagons ZSSK Cargo. ZSSK Cargo. https://www.zscargo.sk
16. Frequentis to support Polish GSM-R deployment. Infrastructure, Europe. Railway Gazette
(2018). https://www.railwaygazette.com
17. Control of road transport: purpose, capabilities and technology. Auto and Moto, Consumer
Guide. KP.RU (2018). https://www.kp.ru
18. Knapčíková, L., Balog, M.: Selected legal conditions valid by transport of dangerous goods
by rail in European Union. Acta Tecnología 3(3), 7–10 (2017). ISSN 2453-675X
19. Alexeev, V.: Wireless GPS/GSM/GPRS systems for remote car diagnostics based on Enfora
modems. Wirel. Technol., 1–12 (2012)
20. GPS/GLONASS vehicle monitoring systems: opportunities, offer market and cost of
solutions. Auto and Moto, Consumer Guide. KP.RU (2016). http://www.kp.ru
21. Private conclusions on the sequence of development of the hitch wagons. http://www.
studopedia.ru
22. Coupling device wagons. Irontrack.ru (2016). http://www.irontrack.ru
23. Container/Tanker/Wagon E-Seal/RFID Rail Monitoring, Rail Cargo Freight Security and
Tracking System, Avante. http://www.avantetech.com
24. MoxaCorporate. Industrial Ethernet Solutions. https://www.moxa.com/solutions/railway/
Solution/
25. GPS monitoring of railway transport, Schturman monitoring. http://gps.shturman.kz
26. Innovation digest JSC RZD. http://www.rzd-expo.ru/innovation/infrastructure/
27. Balog, M., Minďaš, M.: Informatization of rail freight wagon by implementation of the
RFID technology. In: International Summit Smart City 360°, pp. 592–597. Springer, Cham
(2016). ISBN 978-3-319-33680-0
28. Balog, M., Husár, J., Knapčíková, L., Šoltýsová, Z.: Automation monitoring of railway
transit by using RFID technology. Acta Technológia 1, 9–12 (2015)
29. Balog, M., Szilágyi, E., Knapčíková, L.: The identification system of a vertical traffic sign
based on RFID technology is a utility model SK 7695 Y1, Banská Bystrica: ÚPVSR, 4
p. (2017). Original in Slovak
30. Balog, M., Marasová, D., Knapčíková, L., Balog, P.: Intelligent wagon: a new approach to
monitoring the wagon’s technical conditions. In: Kováčiková, T., Buzna, Ľ., Pourhashem,
G., Lugano, G., Cornet, Y., Lugano, N. (eds.) Intelligent Transport Systems – From
Research and Development to the Market Uptake. INTSYS 2017. Lecture Notes of the
Institute for Computer Sciences, Social Informatics and Telecommunications (2018)
31. Balog, M., Straka, M., Šimeková, Ž.: Centre of informatization in rail transport, project of
three organizations. Logistyka 5, 238–241 (2013). ISSN 1231-5478
32. Balog, M., Hricová, R.: Introduction of RFID system into transport and defining its model of
return on investment. Int. J. Eng. Res. Afr. 18, 130–135 (2015). http://www.scopus.com/
record/. ISSN 1663-3571
33. Balog, M., et al.: Experimental verification of RFID system functionality on train kit models.
In: Research Forum 2014: Lecture Book: EU Structural Funds for Research and Innovation -
Effects, Opportunities and Risks, 9–11 April 2014, Grand Hotel Bellevue, Vysoké Tatry.
Výskumné centrum Žilinskej univerzity v Žiline, Žilina, pp. 1–11 (2014). ISBN 978-80-554-
0842-2
34. Autonomous GPS beacon for containers, Intelli. http://intelli.com.ua
35. GPS/GLONASS monitoring railway transport. Wagons Control. http://wagons-control.ua
LiDAR Based System for Tracking Loader
Crane Operator
Karol Miądlicki(&) and Mateusz Saków
Institute of Mechanical Technology,

Faculty of Mechanical Engineering and Mechatronics,
West Pomeranian University of Technology, Szczecin, Poland
karol.miadlicki@zut.edu.pl
Abstract. Nowadays, particularly in the IoT, industry 4.0, smart devices,

intelligent robots and autonomous vehicles, the segmentation and analysis of the
environment using 3D data (point clouds) is a current and relevant research
topic. The reducing prices of depth sensors based on light detection and ranging
(LIDAR), time of flight (ToF), radio detection and ranging (RADAR) tech-
nologies and high computing power, facilitate the development of intelligent
systems. In this work real-time LiDAR based system for tracking loader crane
operator has been presented. Developed system uses a Velodyne VLP-16
LiDAR to detect and track loader crane operator and is fully integrated with
open-source ROS (Robot Operating System). This solution will help to deter-
mine whether the operator is in the crane working zone, which will increase his
safety. The work presents the main assumptions of the system, subsequent
stages of point clouds processing and used algorithms. Detection and tracking
algorithms performance was evaluated based on real data acquired under dif-
ferent conditions, and the results were compared to similar solutions found in the
literature. The results of conducted test were presented. Furthermore, in this
article, we discussed other methods used to detect and track humans in real time,
described Velodyne VLP-16 scanner, robot operating system and data pro-
cessing methods.
Keywords: Work zone segmentation Ground filtration Loader crane

Mobile crane LIDAR VLP-16 RANSAC Pedestrian tracking RoS
Robot operating system Dot product Point-Cloud Real-Time processing
Velodyne
1 Introduction
Due to the increasing availability, decreasing prices, and development of virtual

techniques, three-dimensional (3D) scanners are used in an increasing number of
applications. Scanners based on laser or light beam return time can be divided into two
groups: dense data scanners (high resolution) and sparse data scanners (low resolution).

https://doi.org/10.1007/978-3-030-18715-6_34
LiDAR Based System for Tracking Loader Crane Operator 407
Accurate scanners from, e.g., Riegl and Teledyne Optech generate a highly dense grid
of dots and need much time to perform a full scan. This long scan time makes them
unsuitable for real-time applications. Accurate scanners are used for mapping and site
inspection, creating 3D maps of cities [1], or in forestry [2]. In these applications, a 3D
scan is first performed and then the point cloud is analyzed. The second group is
scanners performing 20–30 scans per second. This group includes Microsoft Kinect v2,
Leap Motion, SICK LMS, Ibeo LIDAR, and Velodyne LIDAR, among others. Despite
the lower resolution of the generated point cloud, the scanning time allows them to be
used in real-time applications. Therefore, the scanners of the second group are com-
monly used in the design of control systems: autonomous cars [3], robots [4] and traffic
inspection [5]. The control systems of these devices are suitable to analyze the envi-
ronment, recognize pedestrians [6, 7] or even find the right paths. In all these appli-
cations, proper segmentation and classification based on sensor data are crucial.
Currently, the loader cranes are commonly controlled by hydraulic levers or control
panels. Joysticks, electronic levers, buttons, switches, light-emitting diodes (LEDs),
and light crystal displays (LCDs) have also become standard. However, following
the trends of dynamically developing industries, together with the Cargotec com-
pany, we decided to develop an innovative control and monitoring system for the
loader crane.
In this paper, an innovative use for the real-time operator tracking algorithm,
developed for low-resolution 3D data obtained from a light detecting and ranging
(LIDAR) sensor, has been presented. The presented algorithm is part of a sensor-fusion
system (Fig. 1) for locating the operator and recognizing his or her gestures that are
applied to control the loader crane. It is an innovative concept, because research on
intelligent loader cranes using extended and virtual reality in their control systems has
been conducted by few research centers. In Szczecin [8–12] researchers presented a
approach to controlling loader crane using exoskeleton. They also used augmented reality
and environment scanning (using stereovision and structural light) to estimate the posi-
tion of the boom. Researchers form Koszalin [13] presented a consistent approach based
on interactive system with voice control for loader cranes. In addition to present the main
principles of the system they focused on intelligent voice interface. Researchers form
Poznan [14, 15] focused on development process of the haptic joystick electronic con-
troller design for loader crane. In Sweden scientist from Umeå University are working on
teleoperation system for a hydraulic crane [16]. In Bryansk in Russia universal mathe-
matical model of a hydraulic loader crane has been developed [17].
408 K. Miądlicki and M. Saków
Fig. 1. Concept of control and monitoring system for loader crane.
At the beginning of the research, together with the Cargotec company two main
problems regarding cranes was identified. The first problem concerns the loader crane
control method. Loader cranes are commonly controlled by hydraulic levers or control
panels. Joysticks, electronic levers, buttons, switches, LEDs and LCD displays also
become standard. All these elements have an impact on: control precision, comfort of
preforming complex movements and setting multiple parameters of the crane without
entering the cabin or going to the control panel. This control methods are complex and
unintuitive. Our goal is to facilitate the way of control, by making it similar to the
control of an industrial robot. It will simplify control and increase operator safety. The
second problem concerns loader crane working area monitoring system, especially in
cities. The purpose of our work is to ensure the possibility of safe operating loader
crane without surrounding the work area with barriers. This goal requires operator
tracking and movement detection system in the working area. Following the trends of
dynamically developing industries, together with Cargotec company we decided to
develop, an innovative safety control and monitoring system for the loader crane.
The tracking algorithm described in this article is a crucial part of a system that uses
a depth scanner to monitor the work area of a loading crane (Fig. 2). The main tasks of
the subsystem are: the segmentation of loader crane work zone, monitoring movements
in this zone and tracking operator position. Subsystem should also facilitate mapping of
the crane surroundings to virtual reality. After analyzing the literature, it was decided to
use a Velodyne VLP-16 laser depth scanner. The main factor determining the choice of
the scanner was usability with real time applications and price. Because the manu-
facturer does not supply any software the raw data received from the sensor requires a
conversion to points in the XYZ space. Then, it is necessary to segment the cloud of
points. Its segmentation is problematic because VLP-16 is the smallest scanner model
offered by Velodyne. Thus, the depth cloud obtained during the scan has a low reso-
lution. Therefore, the algorithm proposed in this paper analyzes sixteen points from
each VLP-16 laser sensor reading. Then, after receiving a full 360-degree scan, the
model of the ground plane is fitted using the RANSAC algorithm. After filtering the
ground points segmentation and classification are performed.
B C
Voice
commands
Following right
Hand hand moves
tracking
control Left hand „start” gesture
Voice „start” command
Fig. 2. Concept of control and monitoring system for loader crane. A – real time operator
tracking, B – unauthorized intrusion detection, C – gestures control
The rest of paper is organized as follows. In the next paragraph, the previous works
on the loader crane interfaces and operator tracking problem has been discussed. Then
the Velodyne VLP-16 scanner and robot operating system has been described. In
Sect. 4 proposed operator tracking system was described. Finally, a conclusion of this
paper and an outline of the future work are given in Sect. 5.
2 Related Work
A detailed overview of the control systems used in loader cranes is presented in [18].
Only a few academic centers are working on the development and testing of control
interfaces and support systems for loader cranes. In [13, 19, 20] the researchers pre-
sented a novel approach to controlling the crane. The authors outlined the main
assumptions of the intelligent system for the loader crane and focused on the devel-
opment of the voice interface. Their system allows the operator to use voice commands
and natural language to control the crane. On the other hand, in [20], the authors
proposed a method of controlling the loading crane using gestures and a Microsoft
Kinect sensor. In 2016, Hiab presented the HiVision control system for forestry cranes.
It allows the operator to monitor the environment around the crane without leaving the
cab. Cameras with a 270° angle of view mounted on the crane arm and Oculus Rift
(virtual reality goggles) were used to implement this system. Research on the use of
self-sensing teleoperation systems to control loader cranes has been described in [21–
27]. Experiments on the use of exoskeletons for crane control has been explained in
[28–32]. A review of the available literature reveals a substantial lack of research on the
crane environment monitoring system or tracking the crane operator’s position using
LIDAR or other vision systems.
3 Hardware and Software
We used the LiDAR VLP-16 sensor to implement our solution. As a hardware platform
for data processing and visualization, we used a PC with a Linux system and a robot
operating system.
3.1 VLP-16 - Velodyne LiDAR Scanner

Laser distance or LIDAR sensors measure distance by illuminating a target with a laser
and analyzing the reflected light. Due to the decreasing price and improving mea-
surement parameters, they have gained more and more attention. LIDAR sensors are
used in many applications: urban planning, urban mapping, intelligent autonomous
transport, and scanning forests and agricultural fields. Used in the described system, the
VLP-16 scanner is the third generation of laser scanners produced by Velodyne.

A scanner is composed of 16 individual laser-detector pairs that are individually aimed
in two-degree increments on the laser head. That sensor configuration provides a 30°
field of view (−15° to 15°), which results in 300,000 depth points per second. Based on
research conducted at the University of Houston [33], where researchers have evaluated
the sensor for long-term stability, geometric calibration, and the effect of temperature
variations, we can conclude that the information about sensor accuracy for the factory
calibration is correct. Moreover, after the local calibration, it is even possible to
improve the sensor accuracy by ±10 mm. The VLP-16 sensor only provides infor-
mation about azimuth (readings from an encoder) and distance (depth). To read the
coordinates of the points (x, y, z) in space, it is necessary to convert the received data
while considering the position of the laser sensor pairs and the current head position
(encoder measurement). Coordinates are calculated based on the simplified equation.
The simplification is allowed because the first six parameters of the equation are
internal sensor calibration constants. These constants are defined by the manufacturer
during the production calibration process of the device at the production stage. Only the
azimuth (encoder reading) and the point distance from the sensor are obtained from the
sensor. The simplified equations for each coordinate and its geometric representation
are shown in Fig. 3. Figure 4 shows the point cloud obtained after converting raw data
from the sensor. Sparse lines visible on the image are caused by only 16 laser sensors in
the VLP-16. Despite the low resolution in Fig. 4, three human bodies, a loader crane,
room walls, and other environmental elements (hydraulic pump, test stand) are visible.
Fig. 3. Raw sensor data to XYZ conversion

Fig. 4. LIDAR scan of the crane test stand: A - side view, B - top view
3.2 ROS – Robot Operating System

ROS is an open-source, meta-operating system. This system contains many tools,
libraries and algorithms and simplifies writing both complex functions and building
entire software packages. Although ROS is not an operating system, it provides services
designed for a complex computation such as hardware abstraction, low-level device
control, implementation of commonly used functionality, message-passing between
processes, and package management. ROS was released undehe BSD license and free
software. It can be used for scientific and commercial purposes for free. The system has
the most support on the Ubuntu platform, however, there are also experimental versions
operating on the following platforms: Debian, Windows, OS X, UDOO.
In described system, we use the ROS system to process and visualize data in real
time. This system simplifies integration entire functions of operator tracking system All
nodes (ellipses), topics (text above links) and the links between them are shown on the
graph (Fig. 5). First data from VLP-16 sensor is received by the topic “/Velo-
dyne_nodelet_manager_driver”. Next data package is distributed to topics
Linux /rviz_nodelet
/show_objects_node
visualizaƟon _manager
/hdl_people
_tracking
People
People
coordinates
classified
points
VLP-16 LiDAR /Velodyne_nodelet /Velodyne_nodelet /hdl_people

/pcl_ground remove
sensor Point _manager_driver Point _manager_cloud Point Point _detecƟon_nodelet
cloud cloud cloud cloud
without
ground
Fig. 5. Simplified ROS computation graph describing the developed system

LiDAR Based System for Tracking Loader Crane Operator
413
“/Velodyne_nodelet_manager_cloud” and “/pcl_ground_remove” (topic that remove

ground points) and next to “/hdl_people_nodelet” (segmentation and classification).
The last data processing topic is “/hdl_tracking_nodelet” where selected sets of point
clouds are tracked. Next data “/velodyne_points” are forwarded to Rviz module topic “/
show_objects_node” for visualization.
4 Tracking System
The operator tracking system consists of three key algorithms: ground filtration, seg-
mentation and classification (human/pedestrian detection). Only the combination of
these three algorithms allows you to determine position and track the loader crane
operator. Data flowchart of entire system was shown below on Fig. 6.
Fig. 6. Algorithms used in loader crane operator tracking system
4.1 Ground Filtration

In the presented filtering solution, we use characteristic scan pattern properties, the dot
product of vectors, and interpolation using the RANSAC method. The algorithm we
propose is detailed described in the following works [34–36]. In our solution we use
scan segmentation, dot product of vectors, and interpolation using the RANSAC
method. In the presented algorithm, we analyze a single scan from 16 laser sensors.
Then, the points are projected onto the YZ plane. Starting with the point having the
smallest Z coordinate value, we create two vectors from three successive points. If the
angle between the vectors DE and EF is zero, the points are probably lying on the
ground plane. To determine this condition, we use the dot product of the DE and EF
vectors, whose value should be close to 1. In the next step, we select points D, F, and
G. The operation is repeated until all points have been analyzed. Because each point is
involved in computing the scalar product of the vectors, we stored all three values in
the array three times. This avoids misclassifying a point if it is near an obstacle or it
there is a slightly uneven terrain, because the threshold is based on two maximum
values from the dot product array. In addition, the use of this matrix allows for the
extraction of the image edges, i.e., the place where obstacles or walls begin. Addi-
tionally, to avoid the detection of walls as a ground plane, the thresholds for differences
in the X coordinate are observed. The last step in the algorithm is to determine the
ground plane. For this purpose, we used the RANSAC method. The linear function is
fitted to the filtered ground points projected on the YZ plane. The segmentation
algorithm is summarized in Algorithm 1 described below.
Algorithm 1: Ground points filtration

Input : IN_points = [x,y,z;….;x16,y16,z16]
Output : [ground_points, ground_model]
: threshold_val, X_tresh, num,iter,
Parameters
threshDist, inlierRatio
1: for i = 0:14
2: vecDE = vector([Y1,Z1],[Yi,Zi])
3: vecEF = vector([Yi,Zi],[Yi+1,Zi+1])
4: cos_th = dot(vecDE vecEF)
5: cos_mem = [cos_thi-2,cos_thi-1,cos_th]
if (sum(cos_mem> threshold_val)>1 &&
6:
Xi-Xi+1 > X_tresh)
7: ground_points = [ground_points,[Xi, Yi, Zi]]
8: Else
9: break
10: end
11: end
12: if full_360_scan == true
ground_model = RANSAC (ground_points,
13:
num,iter, threshDist, inlierRatio)
14: End
4.2 Segmentation and Classification

In the presented segmentation solution, we use modified algorithm developed by
Bogoslavskyi and Stachniss. Used algorithm has been detailly described in [37–39].
Moreover, the principle of its operation was graphically illustrated in Fig. 7a–b, where
the part of scene from Fig. 2 has been analyzed. Figure 7a shows a view of 3 persons
(top-view) with marked laser beams coming from the sensor VLP-16, Fig. 7b explains
a geometrical interpretation of the Eq. 1.
VLP-16
A
human 3
B B
d2
beta
d1
X A
d1
d2
alfa
human 1
human 2
S (VLP-16)
Fig. 7. A, B - graphic algorithm explanation based on the part of scene from Fig. 2, C -
geometric interpretation of VLP-16 polar coordinates to X, Y, Z coordinates conversion
arctankAX k arctanðd2 sinðaÞÞ

b¼ ¼ ð1Þ
kXBk d1 d2 cosðaÞ
Based on the Eq. 1, angle b between points was calculated. That angle determines
if point is a part of object. (label) Threshold value for angle b < 10° was established
during the research. Every point below 10° were assigned to new object. The same
thresholding (based on b angle) can be performed by projecting points on the other two
planes X-Y. For determine neighborhood of the points with the same label’s breadth-
first search method was used. Detection and tracking operator algorithms are based on
the method described by Koide and Miura [40]. After the implementation, tests were
performed, segmentation results are presented in the Fig. 8. Detection and tracking
results are presented in the Fig. 9. The application was run on a PC with a processor i7
4700K and 8 GB of ram memory. On this hardware and poorly optimized code in C++/
Python segmentation and tracking frequency of 50 Hz was achieved. Because the
movement speed of the crane is not high, that value is reasonable.
Fig. 8. Segmentation results side and top view.

Fig. 9. Operator detection and tracking algorithm
5 Conclusions
In this work, we presented a system for loader crane operator tracking using sparse
LIDAR data for loader crane sensor fusion system. The proposed system allows for
accurate tracking operator in real time. Implemented filtration algorithm allows to
accurate filtering ground points from the cloud of points in real time and allows to get a
model of ground plane. Experiments show that combination dot product of vectors and
interpolation using the RANSAC algorithm gives better results than using single
methods. The calculation time for the entire scan is about 20 ms. However, time and
accuracy depend on size memory matrix and RANSAC algorithms parameters.
Developed approach is insensitive to the sensor position changes and does not incor-
rectly classify the wall points.
Tracking algorithm evaluation tests reveal several disadvantages. We have noticed
problems with the segmentation of objects which are close to each other or are close to
the walls. This may be due to the low resolution of the VLP-16 sensor or incorrectly
selected threshold values. After algorithm optimization and applying thresholding
(based of the angle b) in all three dimensions (XYZ), the effectiveness of the method
should increase. Use larger number of sensors or sensors with higher resolution (HDL-
32 or HDL-64) may be a way to improve the system. Another problem consists the low
scanning frequency of the sensor, which makes it impossible to detect fast moving
humans. Next issue concerns the assignment of the same id to the objects that were
detected again. Now a new id is assigned to each detected object.
Despite the disadvantages proposed approach should be regarded as promising. The
results of filtration, segmentation and tracking using proposed algorithm are promising
as shown in Fig. 8. Currently we are working on code optimization and parallelization
with CUDA technology and ROS system. As part of the further work, the imple-
mentation of the wall separation algorithm, complete segmentation of the load crane
environment and cargo tracking algorithm is planned. In addition, it is planned to use
the UNITY engine to transfer the loader crane work zone to virtual reality.
Acknowledgments. The work was carried out as part of PBS3/A6/28/2015 project, “The use of
augmented reality, interactive voice systems and operator interface to control a crane”, financed
by NCBiR.
References
1. Sanjaya, K., Henning, F., Purba, K.R.: 3D LIDAR city model application and marketing
plan development. In: International Conference on Soft Computing, Intelligent System and
Information Technology: Building Intelligence Through IOT and Big Data, pp. 238–242.
IEEE (2018)
2. Gopalakrishnan, R., Thomas, V.A., Wynne, R.H., Coulston, J.W., Fox, T.R.: Shrub
detection using disparate airborne laser scanning acquisitions over varied forest cover types.
Int. J. Remote Sens. 39, 1220–1242 (2018)
3. Zhou, Q., Tan, Z., Yang, C.: Theoretical limit evaluation of ranging accuracy and power for
LiDAR systems in autonomous cars. Opt. Eng. 57, 1–8 (2018)
4. Malavazi, F.B.P., Guyonneau, R., Fasquel, J.B., Lagrange, S., Mercier, F.: LiDAR-only
based navigation algorithm for an autonomous agricultural robot. Comput. Electron. Agric.
154, 71–79 (2018)
5. Sakib, M.N., Rahman, M.A.: Traffic bottleneck reconstruction LIDAR orthoimages: a
RANSAC algorithm feature extraction. Adv. Intell. Syst. Comput. 843, 302–307 (2018)
6. Wu, J., Xu, H., Zheng, Y., Tian, Z.: A novel method of vehicle-pedestrian near-crash
identification with roadside LiDAR data. Accid. Anal. Prev. 121, 238–249 (2018)
7. Nakashima, K., Jung, H., Oto, Y., Iwashita, Y., Kurazume, R., Mozos, O.M.: Learning
geometric and photometric features from panoramic LiDAR scans for outdoor place
categorization. Adv. Robot. 32, 750–765 (2018)
8. Herbin, P., Pajor, M.: The torque control system of exoskeleton ExoArm 7-DOF used in
bilateral teleoperation system. In: AIP Conference Proceedings, pp. 020020–020020-9
(2018)
9. Pajor, M., Grudziński, M., Marchewka, Ł.: Stereovision system for motion tracking and
position error compensation of loading crane. In: AIP Conference Proceedings, pp. 020050–
020050-10 (2018)
10. Jasiewicz, M., Miądlicki, K., Powałka, B.: Assistance of machining parameters selection for
slender tools in CNC control. In: AIP Conference Proceedings, pp. 020024–020024-10
(2018)
11. Jasiewicz, M., Powalka, B.: Identification of a lathe spindle dynamics using extended inverse
receptance coupling. J. Dyn. Syst. Meas. Control 140(12), 121015-1–121015-11 (2018)
12. Jasiewicz, M., Powałka, B.: Prediction of turning stability using receptance coupling. In: AIP
Conference Proceedings, pp. 100005-1–100005-9 (2018)
13. Majewski, M., Kacalak, W., Budniak, Z., Pajor, M.: Interactive control systems for mobile
cranes. Adv. Intell. Syst. Comput. 1, 10–19 (2018)
14. Rybarczyk, D., Owczarek, P., Myszkowski, A.: Development of force feedback controller
for the loader crane. In: Advances in Manufacturing. Lecture Notes in Mechanical
Engineering, pp. 345–354. Springer (2018)
15. Myszkowski, A., Bartkowiak, T., Staniek, R.: Novel type of biaxial wheel-type haptic input
device with MR-brakes to control hydraulic multi-axis manipulators. In: Volume 4A:
Dynamics, Vibration, and Control, pp. 71300-1–71300-6. ASME (2017)
16. Westerberg, S., Manchester, I.R., Mettin, U., Hera, P.L, Shiriaev, A.: Virtual environment
teleoperation of a hydraulic forestry crane. In: IEEE International Conference on Robotics
and Automation, pp. 4049–4054. IEEE (2008)
17. Lagerev, I.A., Lagerev, A.V.: Universal mathematical model of a hydraulic loader crane. In:
IOP Conference Series: Earth and Environmental Science, vol. 194, pp. 032015-1–032015-6
(2018)
18. Miądlicki, K., Pajor, M.: Overview of user interfaces used in load lifting devices. Int. J. Sci.
Eng. Res. 6, 1215–1220 (2015)
19. Majewski, M., Kacalak, W.: Human-machine speech-based interfaces with augmented
reality and interactive systems for controlling mobile cranes. Lecture Notes in Computer
Science, pp. 89–98. Springer (2016)
20. Majewski, M., Kacalak, W.: Intelligent speech-based interactive communication between
mobile cranes and their human operators. Lecture Notes in Computer Science, pp. 523–530.
Springer (2016)
21. Saków, M., Miądlicki, K., Parus, A.: Self-sensing teleoperation system based on 1-DOF
pneumatic manipulator. J. Autom. Mob. Robot. Intell. Syst. 11, 64–76 (2017)
22. Saków, M., Parus, A., Pajor, M., Miądlicki, K.: Nonlinear inverse modeling with signal
prediction in bilateral teleoperation with force-feedback. In: 22nd International Conference
on Methods and Models in Automation and Robotics, pp. 141–146. IEEE (2017)
23. Saków, M., Miądlicki, K.: Transport delay and first order inertia time signal prediction
dedicated to teleoperation. In: Advances in Automation, Robotics and Measurement
Techniques, pp. 142–151. Springer (2018)
24. Saków, M., Parus, A., Pajor, M., Miadlicki, K.: Unilateral hydraulic telemanipulation system
for operation in machining work area. In: Advances in Manufacturing. Lecture Notes in
Mechanical Engineering, pp. 415–425. Springer (2018)
25. Herbin, P., Pajor, M.: Interactive 7-DOF motion controller of the operator arm (ExoArm 7-
DOF). In: International Conference on Innovations in Intelligent Systems and Applications,
pp. 185–188. IEEE (2017)
26. Saków, M., Marchelek, K., Parus, A., Pajor, M., Miądlicki, K.: Signal prediction in bilateral
teleoperation with force-feedback. In: Proceedings in Mathematics and Statistics, pp. 311–
323. Springer (2018)
27. Sakow, M., Marchelek, K.: Model-free and time-constant prediction for closed-loop systems
with time delay. Control Eng. Pract. 81, 1–8 (2018)
28. Herbin, P., Pajor, M.: ExoArm 7-DOF (interactive 7-DOF motion controller of the operator
arm) master device for control of loading crane. In: Advances in Manufacturing. Lecture
Notes in Mechanical Engineering, pp. 439–449. Springer (2018)
29. Herbin, P.: Design and modeling of an exoskeleton torque sensor. Sci. J. Marit. Univ.
Szczecin 51(123), 89–94 (2017)
30. Berczyński, S., Lachowicz, M., Pajor, M.: An improved method of approximating frequency
characteristics in the problem of modal analysis and its applications. Comput. Eng. 30, 575–
584 (2001)
31. Pulikowski, D., Lackner, F., Scheuerlein, C., Meinel, D., Savary, F., Tommasini, D., Pajor,
M.: Testing mechanical behavior of Nb3Sn Rutherford cable during coil winding. IEEE
Trans. Appl. Supercond. 27, 1–5 (2017)
32. Pajor, M., Marchelek, K., Powałka, B.: Experimental verification of method of machine tool-
cutting process system model reduction in face milling. Comput. Methods Exp. Meas. 22,
503–512 (1999)
33. Glennie, C.L., Kusari, A., Facchin, A.: Calibration and stability analysis of the VLP-16 laser
scanner. Int. Arch. Photogram. Remote Sens. Spat. Inf. Sci. - ISPRS Arch. 40, 55–60 (2016)
34. Miądlicki, K., Pajor, M., Saków, M.: Ground plane estimation from sparse LIDAR data for
loader crane sensor fusion system. In: 2nd International Conference on Methods and Models
in Automation and Robotics, pp. 717–722. IEEE (2017)
35. Miadlicki, K., Pajor, M., Saków M.: Real-time ground filtration method for a loader crane
environment monitoring system using sparse LIDAR data. In: International Conference on
Innovations in Intelligent Systems and Applications, pp. 207–212. IEEE (2017)
36. Miadlicki, K., Pajor, M., Saków, M., Loader crane working area monitoring system based on
LIDAR scanner. In: Advances in Manufacturing. Lecture Notes in Mechanical Engineering,
pp. 465–474. Springer (2018)
37. Bogoslavskyi, I., Stachniss, C.: Fast range image-based segmentation of sparse 3D laser
scans for online operation. In: IEEE International Conference on Intelligent Robots and
Systems, pp. 163–169. IEEE (2016)
38. Bogoslavskyi, I., Stachniss, C.: Analyzing the quality of matched 3D point clouds of objects.
In: International Conference on Intelligent Robots and Systems, pp. 6685–6690. IEEE
(2017)
39. Bogoslavskyi, I., Stachniss, C.: Efficient online segmentation for sparse 3D laser scans.
Photogram. Fernerkundung Geoinf. 85, 41–52 (2017)
40. Koide, K., Miura J.: A Portable person behavior measurement system using a 3D LIDAR.
In: The Proceedings of JSME Annual Conference on Robotics and Mechatronics
(Robomec), pp. 11–111. IEEE (2017)
Artificial Neural Network-Supported Selection
of Materials in Ecodesign
Izabela Rojek1 and Ewa Dostatni2(&)

1
Institute of Mechanics and Applied Computer Science,
Kazimierz Wielki University, Bydgoszcz, Poland
2
Department of Management and Production Engineering,
Poznan University of Technology, Poznan, Poland
ewa.dostatni@put.poznan.pl
Abstract. At the current rate of development in the industry, protecting the

global environment is a priority. One of the ways to prevent environmental
degradation is the manufacturing of recyclable products which can be disas-
sembled and their materials and components processed for reuse. Recyclability
of products can be taken into account as early as at the design stage. Appropriate
selection of materials, connections and disassembly methods can facilitate and
simplify recycling. This paper looks at an original method of material selection
at the design stage, which takes into consideration recyclability of materials. An
expert system based on artificial neural networks is presented. Analyses of input
data for material selection have been conducted, the training file and neural
network models have been developed, and the models have been assessed.
Research has proven that artificial neural networks are suitable for supporting
the selection of materials in ecodesign.
Keywords: Material Selection Ecodesign Expert system

Artificial neural network
1 Introduction
The recent years have seen a rapid development of environmental policies and envi-
ronmental awareness. Industrial enterprises are pointed at as one of the main culprits in
environmental pollution. Legal restrictions concerning environmental protection
(standards, statutes, regulations) force enterprises to focus more on environmentally
friendly solutions.
Marketing campaigns aimed at promoting eco-friendly products also shape the
attitudes of manufacturers. Creating customer demand, they force manufacturers to
supply eco-friendly products to the market. Certain customer groups are even willing to
pay a higher price for an environmentally friendly product. Some products in their eco-
friendly versions are cheaper in use. For example, household appliances of higher
energy efficiency class, such as washing machines, use less electricity and help reduce
electricity bills. All the factors mentioned above have induced businesses to introduce
internal environmental policies. One of their components is ecodesign and application
of supporting tools and methods.

https://doi.org/10.1007/978-3-030-18715-6_35
Artificial Neural Network-Supported Selection of Materials in Ecodesign 423
The authors describe an original method supporting designers in the selection of

materials at the design stage to ensure that the product is eco-friendly and recyclable.
The objective of the authors research was to develop expert system being close to
human thinking in terms of the concept of reasoning. Activity of developed expert
system allows to solve certain tasks, like a man who is an expert in the field. The
application of artificial neural networks methods makes possible the creation of such a
system. Artificial neural networks is method of machine learning, which create
knowledge in automatically way, and MLP networks are doing very well with the
classification problem. The described method utilizes artificial neural networks which
facilitate automated selection of materials based on previous experience of the designer.
The proposed method was developed as part of the continued research [1, 2] into
artificial intelligence-based tools supporting ecodesign.
2 Review of Literature
Ecodesign is aimed at making the product more environmentally friendly throughout the
life cycle [3, 4]. It may be considered a design-related strategic activity complying with
the concept of sustainable development. It generates products, services, and knowledge
which facilitate the achievement of sustainable development-related results [5]. From
another standpoint, ecodesign is a proactive approach to management, facilitating the
development of products with reduced environmental impact throughout their life cycle
without compromising their efficiency, functionality, aesthetics, quality or cost [5, 6].
The concept of ecodesign integrates multiple aspects of traditional and environ-
mental design. The ultimate goal is to develop sustainable solutions which meet human
needs [4]. Design no longer only meets functional needs but also creates the company
image and sets directions of environmental protection-oriented development.
Scientists have been exploring ecodesign for many years [7]. The latest research
shows that companies make continuous effort to integrate ecodesign into their business
practice, with various degrees of success [8, 9]. There is ample literature supporting
companies in the implementation of ecodesign and presenting case studies of imple-
mented ecodesign practices [10–12]. However, ecodesign as an approach is still rela-
tively rare.
One of the pillars of ecodesign is product recyclability. Being one of the methods of
environmental protection, recycling is aimed to reduce the consumption of natural
resources and amount of waste [13].
What happens to a product after it is withdrawn from use should be considered as
early as at the initial phases of its life cycle. Recycling is related not only to the last
phase of the product life cycle. Three types of recycling are distinguished:
– recycling in the manufacturing phase (aimed to minimise the consumption of nat-
ural resources and energy),
– recycling during the service life (aimed to extend the service life of products
through the use of replaceable parts),
– recycling at the end of the life cycle (aimed to reuse the materials recovered from
products withdrawn from use) [14].
424 I. Rojek and E. Dostatni
Proper selection of materials for product components is one of the factors which
have an impact on recyclability across the three phases. The design of components must
meet all the functional and strength requirements as well as recycling-related require-
ments specified in the EU legislation and national directives. Therefore, designers need
tools supporting selection of materials at the design stage, taking into consideration
recyclability [15]. Hinchliffe and Akkerman presents the EU Ecodesign and Energy
Labelling Directives, which set regulations on the energy efficiency, environmental, and
information requirements of energy related products entering the EU market [16]. Each
individual product regulation stipulates that the EU Commission must undertake a
review after a certain number of years. This review stage is crucial to improve the
effectiveness and ambition of regulations, and also presents an opportunity to include
new resource efficiency aspects. The strength of a review depends on the review process,
which has followed various approaches in each reviewed product group so far. This
article aims to summarise and discuss these varying review processes and identify their
differences and shortcomings. The analysis is based on comparing the review processes
to the standard Methodology for Ecodesign of Energy related Products (MEErP) used in
preparatory studies to develop new Ecodesign regulations. The article [17] examines the
integration of resource efficiency into the European Ecodesign Directive. A final rec-
ommendation is to continue to expand the work on developing standards defining test
methods and verification procedures on resource efficiency.
This article presents an original method supporting the selection of materials in
ecodesign, using artificial neural networks. In the development of the tool, the authors
have drawn on their previous experience in the development of neural network models
[18, 19] and ecodesign supporting solutions [20, 21].
3 Methods
Selection of materials for product components is supported with the use of neural
network MLP (multilayer feed-forward neural networks with backward error propa-
gation) - one of the methods of machine learning. Artificial neural networks are per-
fectly suited to the acquisition of knowledge and experience, in this case - within the
scope of selection of materials in the ecodesign process.
The construction of a neural network consists in organizing artificial neurons in
several interconnected layers. This is a highly simplified explanation of the brain model
of a living organism, as in fact, the brain does not have a perfectly layered structure
with neutrons interconnected between the layers. Developers of artificial neural net-
works rely on a highly simplified model. Each artificial neural network is built sym-
metrically for easier implementation, and all neurons in one layer are connected to
neurons in the next layer. Redundancy of connections is eliminated in the process of
training the network [22]. A single neuron has a limited capacity. It cannot solve
problems referred to as “linearly non-separable patterns”. Contrary to a single neuron, a
group of neurones, referred to as the artificial neural network (ANN), is perfectly suited
to solving such problems. ANNs have three characteristic features. Neurones are
organised in layers. Each layer can contain a different number of neurones. Layers are
interconnected in such a way that each neuron in one layer is connected to each neuron
in the previous layer. The flow of data is unidirectional, from the input to the output.
Three groups of layers can be distinguished in an MLP:
– the input layer,
– the hidden layer or n- hidden layers, and
– the output layer.
The input layer connects to input variables, normalizes and forwards them to the
hidden layer. Normalization is the process of rescaling input variables to the same
value, e.g., between 0 and 1. Inputs prepared in this way can be forwarded to each
neuron in the hidden layer. The hidden layer is the core layer of an artificial neural
network. An ANN may consist of n- hidden layers. The first hidden layer connects to
the variables in the input layer. The variables are transformed by each neurone in the
hidden layer. Then variables from the first hidden layer are forwarded to subsequent
hidden layers. The last hidden layer forwards the variables to the output layer. The
output layer is built of a number of neurons required to yield a solution [22].
Next, neural network models were deployed in the expert system. An expert system
is computer software which performs complex tasks requiring high intellectual capacity
as competently as a human expert.
Fig. 1. Methodology of application of the neural network model in the expert system for the
selection of materials.
Integration of neural networks into an expert system is a new approach aimed at

utilising the advantages of both. Figure 1 presents an overall methodology of appli-
cation of the neural network model in the expert system for the selection of materials.
4 Expert System Supporting the Selection of Materials

in Ecodesign Using an Artificial Neural Network
4.1 Data Preparation

The process of designing the expert system supporting the selection of materials
encompassed the following stages (Fig. 1):
– input data analysis (on the basis of analyses of material properties),
– development of the training, testing and validating files containing example
selections of materials, to be utilised in the construction of artificial neural networks
and assessment of their efficiency,
– development of artificial neural network models,
– assessment of artificial neural network models,
– development of the expert system supporting selection of materials, with an inte-
grated artificial neural network.
The files were developed on the basis of analyses of material properties, such as the
name of material (e.g. PVC), density expressed in grams per cubic centimetre (e.g.,
7.88), tensile strength expressed in megapascals (e.g., 35.5), yield point elongation Re
expressed in percentage (e.g., 5.5), processing temperature in Celsius degrees (e.g.,
20.8), dielectric constant (e.g., 2.0), dielectric strength expressed in kilowatts per
millimetre (e.g., 22.0), Young’s modulus E expressed in gigapascals (e.g., 4.61), water
absorptivity expressed in percentage (e.g., 22.55), harmful effect on the natural envi-
ronment (e.g., true), recycling cost expressed in PLN per kilogram (e.g., 4.25); a
positive value means profit from the sale of material, a negative value—the cost of
disposal, and the name of additional material (e.g., ABS). Table 1 presents a section of
the training file with example selections of materials. The file with examples contained
980 examples; the testing and validation files—100 examples each.
The file contained both quantitative and qualitative data. The input data processing
stage was applied before being fed into the network input (preprocessing of input data)
and the process of transforming and the appropriate interpretation of the output data
received from the network (postprocessing results) [22]. Min-max normalization was
used to scale the quantitative data. The second category of data are qualitative data that
can be bi- or multi-state, e.g. qualitative has three states: good, limited, incompatible.
One-of-N type coding was chosen which consists in using several variables in the
network structure. For example: for the qualitative number of variables has the value 3
and is equal to the number of possible values of the nominal variable: good = {1, 0, 0},
limited = {0, 1, 0}, incompatible = {0, 0, 1}.
Table 1. A section of the file with examples, from the statistica data mining software.
MM D TS YPE PT DC ME WA RC AM C
ASA 1,04 40 50 160 2,1 400 0,01 100 ABS Good
PA 1,04 35 20 140 2,3 400 0,04 12 ABS Limited
PBT 1,31 29 45 210 2,1 490 0,09 27 ABS Good
SAN 1,08 40 25 180 2,1 500 0,03 40 ASA Good
ABS 1,05 45 30 240 2,2 0 0,1 40 PA Limited
PA 1,04 35 20 140 2,3 400 0,04 12 PA Good
PC 1,20 40 50 210 2,2 40 0,09 80 PA Incompatible
Where: MM – main material, D – density, TS – tensile stress, YPE –
yield point elongation, PT - processing temperature, DC – Dielectric
constant, ME – modules of elasticity, WA – water absorptivity, RC-
recycling cost, AM – added material, C - compability
4.2 Artificial Neural Networks for Selection of Materials

Many models of artificial neural networks were built in experiments. All of them had
the same input layer, whose size was determined by the number of input variables (10
inputs), and the same output layer (1 output). The models were parametrised by various
values. Some of them had one hidden layer but a varied number of neurones in that
layer (from 5 to 25). The number of training epochs in the BFGS algorithm also varied
(from 10 to 120), as well as the error function (SOS or Entropy) and the activation
function in the hidden and output layers (linear, logistic, exponential, Tanh, Softmax).
Table 2 presents the most diverse eight neural networks with various structures, trained
at various parameters.
Table 2. Artificial neural networks for selection of materials.

ID MLP QT QTE QV AT EF HAF OAF
1 10-25-1 96,06 97,96 96,60 38 Entropy Logistic Softmax
2 10-23-1 100,00 100,00 100,00 80 SOS Tanh Linear
3 10-16-1 100,00 100,00 100,00 30 Entropy Tanh Softmax
5 10-12-1 78,43 74,83 76,19 10 Entropy Exponential Softmax
8 10-21-1 100,00 100,00 100,00 60 SOS Logistic Tanh
Where: ID – id network, MLP – name of network, QT - training quality, QTE –
testing quality, QV - validation quality, AT - BFGS training algorithm, ER - error
function, HAF –hidden activation function, OAF – output activation function
The most effective neural networks (100% effectiveness) were MLP 10-23-1, MLP
10-16-1, and MLP 10-21-1. Such good results of the neural network were obtained by
proper selection of the network structure (learning algorithm, error function, activation
function of hidden layer, activation function of output layer and number of neurons in
the hidden layer). A little less effective were MLP 10-25-1 (96.87%), MLP 10-9-1
(97.12%), MLP 10-16-1 (99.22%), and MLP 10-25-1 (96.34%). MLP 10-12-1 turned
out to be the least effective (76.48%). A classification of the artificial neural networks
by effectiveness is presented in Fig. 2.
0.00 25.00 50.00 75.00 100.00
MLP 10-21-1
MLP 10-25-1
MLP 10-12-1
MLP 10-16-1
MLP 10-9-1
MLP 10-16-1
MLP 10-23-1
MLP 10-25-1
validaƟon tesƟng learning
Fig. 2. Effectiveness of artificial neural network models [%].
For an assessment of the artificial neural network models, an error matrix diagram
was developed (Table 3).
Table 3. Neural network error matrix.

Name of network Compatibility - Compatibility - Compatibility -
good incompatible limited
2.MLP 10-25-1-good 265 5 0
2.MLP 10-25-1-incompatible 10 170 15
2.MLP 10-25-1-limited 5 0 510
3.MLP 10-23-1-good 280 0 0
3.MLP 10-23-1-limited 0 0 525
10.MLP 10-12-1-good 225 30 75
10.MLP 10-12-1-limited 45 30 420
Table 3 presents an error matrix for three selected networks. The best network,
MLP 10-23-1, had a 0% rate of errors. MLP 10-25-1 performed a little worse (365
errors), while MLP 10-12-1 showed the poorest performance (220 errors).
4.3 Example Operation of a Material Selection Expert System

The method of material selection using artificial neural networks was implemented into
an expert system. Apart from the selection of materials according to compatibility, the
system performs the selection of an additional element according to compatibility and
the selection of material connections. The expert system can support the designer
and/or process engineer in the selection of materials by eco-friendliness (a detailed
description is included in [2]). In the previous version of the expert system, the
selection of materials according to compatibility was performed based on the machine
learning method. In the current version of the system, the MLP 10-16-1 neural network
is used. The network has proven to be 100% effective in the selection of materials and
is trained over a relatively short time span (30 training cycles), what may be important
when training the network model for new inputs. The error function is Entropy, the
activation function in the hidden layer—Tahn, and the activation function in the output
layer—Softmax.
During selection of materials according to compatibility, the inputs to the expert
system are the parameter values at the input layer of the neural network. The neural
network yields results for the degree of compatibility of materials (e.g. good, limited, or
incompatible).
Selected screenshots from the expert system are presented in Fig. 3. Figure 3a
presents the initial window, Fig. 3b—input of data to the neural network, and Fig. 3c—
the yielded result, i.e., the degree of compatibility (good).
5 Summary
Ecodesign and its practical applications have been a shared domain of interest of
scientists, entrepreneurs and government organisations. Various actions are undertaken
to improve the environmental conditions, from campaigns raising environmental
awareness through legislation to fines for breaching environmental standards. In an
attempt to minimise the environmental footprint, entrepreneurs turn to ecodesign,
supported by the development of methods and tools which facilitate its practical
application. This paper presents one of them—an original, recycling-oriented ecode-
sign supporting method.
Selection of materials for a designed product is a challenging task which requires
considering multiple factors. In the case under analysis, considering a large amount of
numerical input data, it is suitable to use artificial neural networks as the classifying
method. Artificial neural networks show very good classification properties. The
conducted research has proven usefulness and high effectiveness of artificial neural
networks in the support of material selection in ecodesign.
Being highly efficient as data exploration algorithms, artificial neural networks offer
a wide array of possibilities concerning the utilisation of databases. They facilitate
automated acquisition of designers’ knowledge, thus presenting themselves as useful
tools for discovering the knowledge gained by employees through experience.
Fig. 3. Selected screenshots from the expert system supporting selection of materials according
to compatibility.
Acknowledgments. The research work reported here partially was made possible by project no
02/23/DSPB/7716.
References
1. Rojek, I., Dostatni, E., Hamrol, A.: Ecodesign of technological processes with the use of
decision trees method. In: International Joint Conference SOCO 2017-CISIS 2017-ICEUTE
2017, León, Spain, 6–8 September 2017, Proceeding. SOCO 2017, CISIS 2017, ICEUTE
2017. Advances in Intelligent Systems and Computing, vol. 649, pp. 318–327. Springer
(2018)
2. Dostatni, E., Rojek, I., Hamrol, A.: The use of machine learning method in concurrent
ecodesign of products and technological processes. In: Advances in Manufacturing, Lecture
Notes in Mechanical Engineering, pp. 321–330. Springer (2018)
3. Hauschild, M., Jeswiet, J., Alting, L.: Design for environment – do we get the focus right?
Ann. CIRP 53(1), 1–4 (2004)
4. Karlsson, R., Luttropp, C.: Ecodesign: what’s happening? J. Clean. Prod. 14, 1291–1298
(2006)
5. Pigosso, D., Zanette, E.T., Guelere, Filho A., Ometto, A.R., Rozenfeld, H.: Ecodesign
methods focused on remanufacturing. J. Clean. Prod. 18, 21–31 (2010)
6. Johansson, G.: Success factor for integration of ecodesign in product development: a review
of state of the art. Environ. Manag. Health 13(1), 98–107 (2002)
7. Tukker, A., Eder, P., Charter, M., Haag, E., Vercalsteren, A., Wiedmann, T.: Ecodesign: the
state of implementation in Europe. J. Sustain. Prod. Des. 1, 147–161 (2008)
8. Bey, N., Hauschild, M.Z., McAloone, T.C.: Drivers and barriers for implementation of
environmental strategies in manufacturing companies. CIRP Ann. Manuf. Technol. 62, 43–
46 (2013)
9. Dekoninck, E.A., Domingo, L., O’Hare, J.A., Pigosso, D.C.A., Reyes, T., Troussier, N.:
Defining the challenges for ecodesign implementation in companies: development and
consolidation of a framework. J. Clean. Prod. 135, 410–425 (2016)
10. Domingo, L., Buckingham, M., Dekoninck, E., Cornwell, H.: The importance of
understanding the business context when planning ecodesign activities. J. Ind. Prod. Eng.
32, 24–32 (2015)
11. Knight, P., Jenkins, J.O.: Adopting and applying ecodesign techniques: a practitioners
perspective. J. Clean. Prod. 17, 549–558 (2009)
12. Karwasz, A., Trojanowska, J.: Using CAD 3D system in ecodesign-case study. In: Efficiency
in Sustainable Supply Chain, Book Series: EcoProduction, pp. 137–160 (2017)
13. https://pl.wikipedia.org/wiki/Recykling. Accessed Nov 2018
14. Korzeń, Z.: Ecologistics, [in Polish], Biblioteka Logistyka, Poznań (2001). (The Institute of
Logistics and Warehousing)
15. Birch, A., Hon, K.K.B., Shor, T.: Structure and output mechanisms in design for
environment (DfE) tools. J. Clean. Prod. 35, 50–58 (2012)
16. Hinchliffe, D., Akkerman, F.: Assessing the review process of EU ecodesign regulations.
J. Clean. Prod. 168, 1603–1613 (2017)
17. Bundgaard, A.M., Mosgaard, M.A., Remmen, A.: From energy efficiency towards resource
efficiency within the ecodesign directive. J. Clean. Prod. 144, 358–374 (2017)
18. Rojek, I.: Neural networks as prediction models for water intake in water supply system.
Lecture Notes in Artificial Intelligence, LNAI, vol. 5097, pp. 1109–1119. Springer (2008)
19. Rojek, I.: Neural networks as performance improvement models in intelligent CAPP
systems. Control Cybern. 39(1), 55–68 (2010)
20. Dostatni, E., Diakun, J., Grajewski, D., Wichniarek, R., Karwasz, A.: Multi-agent system to
support decision-making process in design for recycling. Soft. Comput. 20(11), 4347–4361
(2016)
21. Dostatni, E., Diakun, J., Grajewski, D., Wichniarek, R., Karwasz, A.: Functionality
assessment of ecodesign support system. Manag. Prod. Eng. Rev. 6(1), 10–15 (2015)
22. Tadeusiewicz, R., Chaki, R., Chaki, N.: Exploring Neural Networks with C#. CRC Press
Taylor & Francis Group, Boca Raton (2014)
The Application of Artificial Neural Network
in Diagnosis of Sleep Apnea Syndrome
Renata Ferduła1, Tomasz Walczak1(&), and Szczepan Cofta2

1
Institute of Applied Mechanics, Poznan University of Technology,
Poznań, Poland
tomasz.walczak@put.poznan.pl
2
Department of Pulmonology, Allergology and Respiratory Oncology,
Poznan University of Medical Sciences, Poznań, Poland
Abstract. In this paper authors propose a method of sleep analysis based on the
algorithm of artificial neural networks. Unlike polysomnography methods, that
are used commonly in clinical practice, the described method does not require
specialist equipment or a qualified technician to analyze biomedical signals. The
results presented in this work show that the properly implemented neural net-
work algorithm can determine incidents during sleep and recognize its phases.
Main idea was tested on the base of data collected from sleep laboratory of eight
patients. From many signals collected during clinical assessment only two were
taken under further consideration: heart rate and blood saturation. As it was
shown, these two parameters measured during sleep allows to determine inci-
dents occurring during sleeping and even to recognize actual stage of sleep. It
means, that it is possible to use simple device that measures only heart rate and
blood saturation to identify sleep apnea syndrome. The method is very effective
and can replace the existing ways to recognizing sleep problems especially,
when sleep examination of patient is conducted in home conditions.
Keywords: Sleep apnea syndrome Apnea-hypopnea index Sleep stages

Artificial neural networks Home sleep apnea testing
1 Introduction
Good sleep is essential for our life and health. Lack of sleep causes difficulties with
concentration, serious behavioral and cognitive changes, irritation and lapses of
memory. Permanent lack of sleep can even end with depression and probably death
(deprived of sleep rats died in four weeks) [1]. Taking care of sleep is very important
and affects all areas of life. So quick diagnosing of sleep disorders is essential [1, 2].
1.1 Sleep Apnea Syndrome

Sleep apnea syndrome (SAS) is defined as repetitive episodes of inability to breath
during sleep [1]. It is caused by relaxation of upper airway muscles during sleeping.
This narrow or closes the airway and makes breathing impossible. As the oxygenation
of blood decreases, patients awake, makes a few deep breaths and falls asleep again.

https://doi.org/10.1007/978-3-030-18715-6_36
The Application of ANN in Diagnosis of Sleep Apnea Syndrome 433
SAS is commonly measured in form of AHI (apnea-hypopnea index) which means

number of apneas and hypopneas per hour of sleeping. Apnea means total blockage of
breathing (shortness of breath is lower than 10% of baseline amplitude) and hypopnea
means shortness of breath not lower than 50% in the presence of reduced oxygenation
of blood for at least 4% compared to state before this episode. Each of these (apnea and
hypopnea) must last for at least 10 s and cause desaturation about 3%. Making medical
diagnosis low blood oxygenation during night and level of daily symptoms are also
taken under consideration. Three types of SAS are distinguished: central, obstructive
and mixed. The obstructive SAS is more common [1, 2].
Considering AHI, SAS can be divided into 3 levels:
– mild: from 5 to 14 events per hour of sleep,
– moderate: from 15 to 29 events per hour of sleep,
– serve: 30 and more events per hour of sleep [3].
Even if AHI is low, serve SAS can be diagnosed because of small duration of deep
sleep (less than 5%) or because of significant desaturation (less than 90% of oxygen
saturation lasting for more than 10% of sleep) [2].
1–2% of children, 2–15% of adult and more than 20% of elder people suffer from
SAS and these numbers are constantly rising. Among others it is because of growing
obesity which is one of most common reason of this disease. Also, knowledge about it
still grows and people earlier notice the symptoms. For this reason, more and more
people go to Sleep Laboratories. In many countries people with clinical symptoms of
SAS are waiting months for assessment of sleep (in United Kingdom in some locations
they have to wait up to 60 months) [4].
In-laboratory polysomnographic examination is recommended by the American
Academy of Sleep Medicine because this method was well tested, and its failure rate is
low. But it has many disadvantages: it is expensive, requires assistance of well qualified
crew, and analyzing results takes much time. Patients have limited access to it and have
to wait for examination and results. It is also important, that polysomnography is
examination with use of many sensors which is uncomfortable and stressful for
patients. The alternative are devices called home sleep apnea testing (HSAT). Such
devices are smaller, cheaper, more comfortable for patient and the examination can be
conducted in patient’s home. But such devices record only some of polysomnographic
signals. Enough to recognize apneas and hypopneas, but not enough to recognize sleep
stages or even estimate sleep duration [5]. This prevent measuring total sleep time
(TST), only letting to measure total recording time (TRT). The lack of information
about time of sleeping makes impossible to count AHI, which means number of apneas
and hypopneas per hour of sleep. Therefore the American Academy of Sleep Medicine
(AASM) recommends using respiratory event index (REI) rather than AHI. For this
reason, there is counted number of apneas and hypopneas per hour of examination and
the differences between these numbers and AHI depend on relationship between
sleeping time and time of not sleeping. Then HSAT can decreases the result of the
severity of SAS. Also HSAT does not allow to control sleep stages (especially deep
sleep) [6–10].
434 R. Ferduła et al.
The authors of this study decided to create and evaluate artificial neural network
model to find easier and cheaper way of diagnosing SAS but allowing the analyze not
only apnea and hypopnea events but also sleep stages. The aim was also to make
analyze of record automatically, and spare of the technician’s time. From many
polysomnographic signals oxygenation of blood and heart rate were chosen because
they are simple to measure. The aim of this study was also to show how useful can
artificial neural network be in such diagnosis and to check possibility of cheaper and
easier access to SAS examination.
1.2 Sleep Study

Sleep medicine started to develop when polysomnography began to be used. Overnight
observations especially electroencephalography (EEG), allowed scientists to learn more
about sleep physiology [11].
Fig. 1. Example of sleep architecture of a healthy man: Wk – not sleeping, REM – REM phase,
S1 – stage 1 NREM, S2 – stage 2 NREM, S3 – stage 3 NREM, S4 – stage 4 NREM [based on
14]
When scientist started long-term sleep studies, they noticed a predictable sequence
of patterns. They separated two states in sleep: REM (rapid eye movement) and NREM
(Non-REM). NREM state has been divided into four stages, where stage 1 means the
lightest sleep and stage 4 – the deepest one. Over time stage 3 and 4 were connected
and now three stages of NREM state of sleep can be considered.
Sleep starts with stage 1, goes quite rapid through stage 2 to deep sleep, which lasts
about 30 min. It is marked by slowing of electroencephalogram. Then it goes back to
stage 2 and 1, and then to REM. This sequence lasts about 90–100 min and repeats a
few times during the night [11, 12].
Sleep stages are evaluated in epoch (lasting 30 s) on the basis of electroen-

cephalographic, oculographic and electromyographic record. The record of stages of
sleep is called sleep architecture (Fig. 1). As time passes, deep sleep is reducing, and
REM is increasing. During all night the largest share is stage 2 NREM (45%–55% time
of sleeping). REM is about 20% time of sleeping and stage 3 NREM – from 10% to
20%. The shortest one is stage 1 NREM (5%–10%) [12–14].
Polysomnography consists of EEG (electroencephalography), EOG (electroocu-
lography), EMG (electromyography), ECG (electrocardiography), thoracic and
abdominal respiratory movements, air flow through upper airway, pulse and oximetry.
The examination should last for at least 6 h of the night [12].
2 Research Problem
The primary objective of this study was to find solution for easier diagnosing OBS with
the use of Artificial Neural Network. Eight patients participated in this research whose
symptoms suggested severe obstructive sleep apnea syndrome (daytime sleepiness,
loud snoring, waking with a dry mouth, falling asleep during ritual activities, attention
problems). Experimental procedures were carried out in University Hospital of Lord’s
Transfiguration in Sleep Laboratory. Patients were examined by a doctor and prepare
por polysomnographic examination by technician. All patients were examined using
polysomnography for one night for at least six hours. Lot of attention was given to
letting patients feel comfortable. The temperature was matched to their preferences and
there was dark whole night. There were at least 7 EEG electrodes: two occipitals, two
central and two parietals, and also one reference electrode. There were also two ECG
electrodes and three EOG electrodes. The thoracic and abdominal respiratory move-
ments, nasal pressure, pulse and oxygenation of blood were also recorded. A technician
analyzed results of the polysomnographic examinations and marked sleep stages,
desaturation, hypopneas and apneas. The parameters were classified according to the
standards.
The record from all sensors can be observe and analyzed using RemLogic. Data
form sensors (heart rate, oxygenation of blood and marked by technician sleep stages,
apnea and hypopnea) were exported as text files in order to further processing using
MATLAB.
In the aim of determining all considered sleep parameters the artificial neural
networks were introduced. Algorithms based on ANN are perfect tools for seeking
relationships between collected sets of data, in which these relationships are compli-
cated, strongly nonlinear or just unknown. In bioscience, ANNs were successfully
applied into identification, interpolation or recognition tasks [15–17].
From obtained data only heart rate, oxygenation of blood, marked sleep stages,
apnea and hypopnea were taken for further analyze. The frequency was reduced to
2 Hz. Then data were prepared as inputs and outputs for artificial neural network to
check when apnea and hypopnea occur and then whether patient sleeps or not and to
mark stages of sleeping. Information about breathing was changed into binary. Apneas
and hypopneas were marked as 1 and normal breathing as 0. Sleep stages were marked
as 1, 2 and 3 for NREM stages, 0 for REM and −1 for not sleeping. Heart rate varied
from 40 to 130 and oxygenation – from 45 to 100 expressed by means of a whole

number. The research was divided into three parts. First recognizing by the ANN
apneas and hypopneas. Then recognizing by the ANN sleep stages. And the last step
was counting duration of sleep and each sleep stage, AHI and desaturation.
There was used neural network for time series prediction from MATLAB Neural
Network Toolbox. The authors chose nonlinear autoregressive neural network with
external input, because in such signals each sample depends on earlier ones. It was
predicting output on the basis of current input and ten past values of input data
(oxygenation of blood and heart rate) and ten past values of output data: stages of sleep
(Fig. 2) and apneas and hypopneas (Fig. 3) which gives 32 input neurons. Some
examples of input and output data are shown in Tables 1 and 2. There were ten hidden
neurons. The network was trained using Levenberg-Marquard algorithm, where 70% of
input data were used to train network, 15% for validation and 15% for testing. Number
of samples oscillated between 240 and 260 thousand. Input matrices which were used
to teach ANN contained data of seven patients in three columns: for recognizing sleep
stages heart rate, oxygenation and sleep stage and for recognizing apneas and
hypopneas heart rate, oxygenation and information about breathing. Then the ANN
was tested. The matrix containing two columns (heart rate and oxygenation of eighth
patient) was the input and the matrix containing one column (sleep stage or information
about breathing) was the output.
There were eight tests conducted. Each time data from seven patients were used to
train network and it was tested on the basis of data from eighth one.
Fig. 2. The architecture of modeled artificial neural network for recognizing breathing pattern (x
– heart rate and blood oxygenation, y – sleep stage)
Table 1. An example of input and output data for recognizing breathing pattern
Samples 1 2 3 4
Input Heart rate x(t) 66 66 66 66
x(t − 1) 66 66 66 66
x(t − 2) 66 66 66 66
… … … … …
x(t − 10) 64 65 65 65
Oxygenation x(t) 89 89 89 89
x(t − 1) 89 89 89 89
x(t − 2) 89 89 89 89
… … … … …
x(t − 10) 91 90 90 90
Breathing pattern y(t − 1) 0 0 0 0
y(t − 2) 0 0 0 0
… … … … …
y(t − 10) 0 0 0 0
Output Breathing pattern y(t) 0 1 1 1
Fig. 3. The architecture of modeled artificial neural network for recognizing sleep stages (x –
heart rate and blood oxygenation, y – apnea-hypopnea or normal breathing)
Table 2. An example of input and output data for recognizing sleep stages
Samples 1 2 3 4
Input Heart rate x(t) 73 73 73 73
x(t − 1) 72 73 73 73
x(t − 2) 71 72 73 73
… … … … …
x(t − 10) 70 70 70 71
Oxygenation x(t) 93 93 93 93
x(t − 1) 94 93 93 93
x(t − 2) 94 94 93 93
… … … … …
x(t − 10) 94 94 94 94
Sleep stage y(t − 1) −1 −1 −1 −1
y(t − 2) −1 −1 −1 −1
… … … … …
y(t − 10) −1 −1 −1 −1
Output Sleep stage y(t) −1 −1 −1 −1
3 Results
The results of learning of ANN are shown in Tables 3 and 5 (firstly sleep examination
and secondly study of breathing pattern). There were calculated mean squared errors
(MSE), number of wrong marked samples and relative errors (the last two parameters
refer to testing ANN on eights patient’s data).
The results of testing ANN are shown in Tables 4 and 6 (Table 4 deals with sleep
examination and Table 6 – breathing pattern). In the Table 4 are results of the calcu-
lations of durations of sleep (TST) on the basis of ANN results and sleep stages marked
by technician. In this table is also presented total recording time (TRT) and durations of
sleep stages in hours and percentage of TST. Each duration was calculated as a sum of
differences between the end of each stage and the beginning if this stage. In the Table 6
number of marked apneas and hypopneas and AHI calculated based on ANN and
technician are compared. Number of apneas and hypopneas was a number of breath
events lasting at least 10 s. AHI was calculated as a quotient of number of breathing
events and total sleep time in hours. AHI was also compered with number of apneas
and hypopneas divided by total recording time (number of apneas and hypopneas per
hour of examination). In the end there is shown duration of desaturation (blood oxy-
genation lower than 90%) expressed in hours and percentage of TST. As it is shown in
Figs. 4 and 5, errors occur only in instants when the state was changing – neural
network was always one sample late. So, duration of apneas, hypopneas and sleep
stages marked by technician and neural network was the same.
Table 3. Artificial neural network parameters after sleep examination

ANN parameters p1 p2 p3 p4 p5 p6 p7 p8
Training MSE [10−4] 5,54 5,96 5,91 5,57 5,58 4,50 5,65 6,30
Validation MSE [10−4] 6,66 4,00 4,54 5,16 6,47 3,93 7,14 4,65
Testing MSE [10−4] 5,56 4,77 4,30 4,32 4,01 15,65 2,50 2,40
Number of errors 60 57 79 61 136 141 20 39
Relative error [%] 0,11 0,11 0,15 0,11 0,25 0,29 0,05 0,09
Table 4. Results of sleep examination by artificial neural network trained (a-h – apnea or
hypopnea, TST – total sleep time, TRT - total recording time)
Durations p1 p2 p3 p4 p5 p6 p7 p8
TST by network [h] 6,39 6,04 6,07 6,53 6,73 4,54 4,29 5,68
TST by technician [h] 6,39 6,04 6,07 6,53 6,73 4,54 4,29 5,68
TRT [h] 7,49 7,57 7,12 7,72 7,46 6,74 5,63 6,36
Time of deep sleeping [h] 1,64 1,20 2,24 1,24 1,70 2,91 0,13 0,69
Time of deep sleeping [%] 25,70 19,90 37,0 19,00 25,30 64,20 3,10 12,20
Time of stage 1 NREM [h] 1,30 1,51 0,93 1,24 0,66 0,49 0,92 0,75
Time of stage 1 NREM [%] 20,40 25,00 15,40 19,00 9,90 10,80 21,60 12,80
Time of stage 2 NREM [h] 3,13 1,35 1,90 2,54 1,53 1,09 3,23 3,69
Time of stage 2 NREM [%] 49,00 22,30 31,30 38,90 22,80 23,90 75,30 65,10
Time of stage REM [h] 0,31 1,98 0,99 1,51 2,82 0,05 0,00 0,57
Time of stage REM [%] 4,90 32,80 16,40 23,10 42,00 1,10 0,00 10,00
Fig. 4. An example of sleep architecture with marked wrong answers of ANN (−1 – not
sleeping, 0 – REM, 1, 2, 3 – 1, 2 and 3 stage of NREM)
Table 5. Artificial neural network parameters after sleep examination

ANN parameters p1 p2 p3 p4 p5 p6 p7 p8
Training MSE [10−2] 0,96 1,01 1,08 1,05 0,78 1,02 1,12 1,02
Validation MSE [10−2] 0,93 1,10 1,06 1,06 0,79 1,09 1,09 1,09
Testing MSE [10−3] 14,60 9,40 5,13 6,95 25,84 7,85 0,14 8,40
Number of errors 804 522 264 388 1408 386 4 390
Relative error [%] 1,49 0,96 0,51 0,70 2,62 0,80 0,01 0,85
Table 6. Results of study of breathing pattern by artificial neural network trained (a-h – apnea
or hypopnea)
Breathing p1 p2 p3 p4 p5 p6 p7 p8
pattern’s
parameters
Number of a-h 402 261 132 194 700 193 2 193
marked by
network
AHI marked by 62,95 43,21 21,76 29,71 104,02 42,47 0,47 34,00
network
Number of a-h 401 261 132 194 700 193 2 193
marked by
technician
AHI marked by 62,95 43,21 21,76 29,71 104,02 42,47 0,47 34,00
technician
Number of a-h 53,68 34,50 18,53 25,14 93,81 28,64 0,36 30,32
per hour of
examination
Time of 2,61 0,19 0,16 6,02 6,40 2,52 0,37 0,98
desaturation [h]
Time of 41 3 3 92 95 55 9 17
desaturation [%]
Fig. 5. An example of apnea and hypopnea diagram with marked wrong answers of ANN
In Table 7 there are shown some of obtained results and evaluation of the exam-
ination which could be prepared as a result of polysomnographic examination.
Table 7. Evaluation of the results (a-h – apnea or hypopnea, TST – total sleep time, TRT – total
recording time, M – moderate SAS, S – serve SAS)
SAS parameters p1 p2 p3 p4 p5 p6 p7 p8
TST [h] 6,39 6,04 6,07 6,53 6,73 4,54 4,29 5,68
TRT [h] 7,49 7,57 7,12 7,72 7,46 6,74 5,63 6,36
TST [%] 26 20 37 19 25 64 3 12
Time of stage 1 NREM [%] 20 25 15 19 10 11 22 13
Time of stage 2 NREM [%] 49 22 31 39 23 24 75 65
Time of stage REM [%] 5 33 16 23 42 1 0 10
AHI 63,0 43,2 21,8 29,7 104,0 42,5 0,5 34,0
Number of a-h per hour of examination 53,7 34,5 18,5 25,1 93,8 28,6 0,4 30,3
Time of desaturation [%] 41 3 3 92 95 55 9 17
Evaluation on the basis of:
AHI S S M M S S – S
Number of a-h per hour of examination S S M M S M – S
Deep sleeping – – – – – – S –
Desaturation S – – S S S – S
The evaluation was prepared on the basis of four components: AHI, number of
apneas and hypopneas per hour of the examination, desaturation and duration of deep
sleeping (stage 3 NREM). S means serve SAS and M – moderate SAS.
4 Discussion
The first thing that has been noticed is that the results of the network are incorrect only
in the instance of change - they are always delayed by one sample. Other than that
delay, the network is doing flawlessly, even though the scholar has been on such a
small group (only 8 patients, and every person has a different basic heart rate and
oxygenation). This raises high hopes for the use of this method in SAS diagnostics and
the replacement of polysomnographic examination. Earlier, however, it will be nec-
essary to create a much larger database of patients examined and to take into account
their accompanying diseases.
As it is shown in the Table 7, sometimes it makes a difference whether number of
apneas and hypopneas is referred to duration of sleeping or to duration of examination,
but among these eight patients the difference wasn’t big. Choose between sleep
duration or examination duration only changed the diagnosis between moderate and
serve SAS for patient 6. If great desaturation is taken into account, it can also some-
times change the diagnosis like in the case of fourth patient.
Far greater difference in diagnosis could be noticed by observing duration of deep

sleeping. Based on AHI only, technician could say that the seventh patient is healthy.
But when sleep architecture is taken into consideration, one can see that this patient had
no REM and very little deep sleep. There could be two different reasons. Maybe the
patient was stressed about the examination and many sensors that his sleep was light (in
this case the technician doesn’t know, whether the patient would have more apneas and
hypopneas if he was relaxed). But perhaps the patient had small hypopneas or sensor
wasn’t properly mounted. With polysomnographic examination it would be necessary
to repeat it. Using HSAT the problem would be unnoticed. Examination with use of
only oximeter to record heart rate and oxygenation of blood could allow to eliminate
situation of not properly mounted sensor. It also can decreases stress of patient what
leads to better quality of records of the sleep stages. And using ANN duration of sleep
stages can be still monitored.
5 Conclusions
The artificial neural networks can be useful in diagnosing Sleep Apnea Syndrome. As it
was shown in this study, polysomnographic examination isn’t necessary, because
results as good as from polysomnography can be achieved from oximeter. This cheap
device is easy to use so everyone could operate it even without instructions. Another
advantage is that no one have to oversee the course of the examination and in opposite
to polysomnography, sensors are rather in fixed places and the possibility of detaching
device is small. So, the examination can be executed at patient’s home. In conjunction
with previous and lower number of sensors patient is less stressed so results are more
reliable. Presented solution is better than other home sleep apnea testing devices
because sleep stages can be recognized which can drastically change the diagnosis.
Thanks to neural networks this solution is better and cheaper than hospitals solution
because results are generated automatically without specialized crew. As it was shown
in this research ANNs are could improve SAS examination. Oximeters could be used in
hospitals and be lend to patients with suspicion of SAS to use at home for a few days.
They could also be in GP doctors’ offices also to lend patients with SAS symptoms.
This solution can be also used in small and cheap devices for people who want to
monitor their sleep.
Acknowledgments. The presented research results were funded with the grant 02/21/DSPB/
3513 allocated by the Ministry of Science and Higher Education in Poland.
References
1. Caldwell, J.P.: Sleep: The Complete Guide to Sleep Disorders and a Better Night’s
Sleep. Firey Books (1997)
2. American Psychiatric Association, Sleep wake disorders DSM-5 Selections. American
Psychiatric Association (2015)
3. Cofta, S.: The assessment of the influence of the degree of obstructive sleep apnea severity
on the chosen parameters of pathogenic processes intensity including selectins and oxidative
stress (2013). (in Polish). (Wydawnictwo Naukowe Uniwersytetu Medycznego im. Karola
Marcinkowskiego w Poznaniu)
4. Flemons, W.W., Douglas, N.J., Kuna, S.T., et al.: Access to diagnosis and treatment of
patients with suspected sleep apnea. Am. J. Respir. Crit. Care Med. 169, 668–672 (2004)
5. Claman, D., Sunwoo, B.: Improving accuracy of home sleep apnea testing. J. Clin. Sleep
Med. 13, 9–10 (2017)
6. Bianchi, M.T., Goparaju, B.: Potential underestimation of sleep apnea severity by at-home
kits: rescoring in-laboratory polysomnography without sleep stagings. J. Clin. Sleep Med.
13, 551–555 (2017)
7. Kapur, V.K., Auckley, D.H., Chowdhuri, S., et al.: Clinical practice guideline for diagnostic
testing for adult obstructive sleep apnea: an American Academy of Sleep Medicine clinical
practice guideline. J. Clin. Sleep Med. 13, 479–504 (2017)
8. Collop, N.A., McDowell Anderson, W., Boehlecke, B., et al.: Clinical guide-lines for the use
of unattended portable monitors in the diagnosis of obstructive sleep apnea in adult patients.
J. Clin. Sleep Med. 3, 737–747 (2007)
9. Epstein, L.J., Kristo, D., Strollo, P.J., et al.: Clinical guideline for the evaluation,
management and long-term care of obstructive sleep apnea in adults. J. Clin. Sleep Med. 5,
263–276 (2009)
10. Setty, A.R.: Underestimation of sleep apnea with home sleep apnea testing compared to in-
laboratory sleep testing. J. Clin. Sleep Med. 13, 531–532 (2017)
11. Madhu, K.: Neurobiology of sleep. 55(Suppl. 2), 10 (2006)
12. Fairbanks, D.N.F., Fujita, S., Ikematsu, T.M.D., Simmons, F.B.: Snoring and Obstructive
Sleep Apnea. Raven Press (2016)
13. Kalia, M.: Neurobiology of sleep. Metab. Clin. Exp. 55(Suppl. 2), S2–S6 (2006)
14. Koziej, M., Mańkowski, M., Zeiliński, J.: Breathing disorders during sleep. Warszawa
(1997). (in Polish). (Wydawnictwo Lekarskie PZWL)
15. Walczak, T., Grabski, J.K., Szadkowska, D., Michałowska, M.: Application of artificial
neural networks in the human identification based on thermal image of hands. In:
Proceedings of the International Conference of the Polish Society of Biomechanics,
Biomechanics in Medicine and Biology. Springer (2019)
16. Grabski, J.K., Walczak, T., Michałowska, M., Cieślak, M.: Gender recognition using
artificial neural networks and data coming from force plates. In: Proceedings of the
Conference on Innovations in Biomedical Engineering. Advances in Intelligent Systems and
Computing. Springer (2018)
17. Michałowska, M., Walczak, T., Grabski, J.K., Grygorowicz, M.: Artificial neural networks
in knee injury risk evaluation among professional football players. In: Proceedings of the
22nd International Conference on Computer Methods in Mechanics (2018)
Artificial Hand Controlled by a Glove
with a Force Feedback
Marcin Białek(&), Dominik Rybarczyk, Andrzej Milecki,

and Patryk Nowak
Department of the Mechatronics Devices, Poznan University of Technology,

Jana Pawła II 24, 61-138 Poznań, Poland
marcin.bialek@student.put.poznan.pl
Abstract. This article describes a construction and control of the hand pros-
thesis with a force feedback. Introduction part contains a review of literature
within the scope of characteristics of hand prosthesis’ designs. There has been
presented different approaches used during creation of artificial hands including
diversity of force feedbacks and mechanisms. To control the artificial hand there
has been used a glove with flex sensors and vibration motors. Such construction
allows to bend the digits with instant tactile sensing. Vibration motors receive
signals provided by force sensing resistors (FSR) placed on phalanges of
prosthesis’ fingers. The movement of artificial hand’s digits is dispensed by
four-bar linkage mechanism. Based on literature overview and performed
research, last part presents the conclusions about the solution of present prob-
lems related to the process of hand prosthesis’ design with special attention to
the aspect of tactile sensing providing and fingers’ kinematic.
Keywords: Hand prosthesis Four-bar linkage mechanism Teleoperation

Force sensing resistors Vibration motors FDM
1 Introduction
The prosthesis is usually built to substitute a part of a body, which in most cases are
limbs or legs. Considering functionality issue, prosthesis can be divided into two
categories: cosmetic and mechanistic. Thanks to ceaseless progress in development of
materials and control methods, since the last years the requirements for prosthesis
increased significantly. It relates to possibilities of replicating real organs and limbs by
artificial prosthesis. Design of the hand prosthesis needs to be similar to its natural
correspondent. The human hand consists of tendons, muscles and skin receptors. Those
components are essential due to their involvement in the process of movement of each
digit and sensory reception of a stimulus. The major issue in the prosthesis design is to
recreate the reflection of human hand’s kinematic complexity. A simple solution is
based on pulley system [1, 2] taken from a biomechanics of elbow joint. A tenet of arm
diffracting and straightening is realized by two muscles (biceps, triceps) working
contrary. A movement of digit is different because muscles that they use are hooked in
five spots (Fig. 1).

https://doi.org/10.1007/978-3-030-18715-6_37
Artificial Hand Controlled by a Glove with a Force Feedback 445
Fig. 1. A – Cyborg Beast [1], B – Bionic Hand Prosthesis [2], C – UTAH/M.I.T. Dextrous
Hand [4], D – Gifu Hand II [9].
The movement of human fingers is being activated by tendons [3]. The single finger
(except thumb) consist of two kinematic pairs class V (between phalanges) and one
kinematic pair class IV (between the basal phalanx and the metacarpal bone). Real-
ization of kinematic pair class IV within hand’s density would be difficult and
unnecessary in basic prosthesis applications. For this reason, it is usually simplified to
kinematic pair class V. This allows the movement of the finger to act as a pulley system
that can use rollers and ropes [4]. Natural thumb has five degrees of freedom
(DoF) provided by carpometacarpal joint (two orthogonal axis), metacarpophalangeal
joint (two orthogonal axis) and interphalangeal joint (single axis) [5]. The average grip
force for a healthy hand is 300 N [6].
Articles about prosthesis describe also constructions that focus on reducing the
number of drives by launching complex mechanisms [7]. They allow operation in more
than one plane (axis) increasing the usability of the device, while simultaneously
subtracting geometry and additional mass of motors. The touch and pressure receptors
allow to perform a proper selection of the pressure applied by the fingers on the
manipulation object. Thanks to this, it is possible to grab objects without the risk of
damaging them [8]. In the case of a human hand, this task is performed by sensory
446 M. Białek et al.
receptors and nerves. They allow proper coupling of the manipulation process. In
prostheses they must be replaced by different sensors. Typically, pressure sensors [9,
10] are used as feedback in such applications.
Serious obstacle in design of hand prosthesis is how to perform communication and
to obtain the appropriate control signals. One of the solutions to this problem is
electromyography (EMG) which records muscles activity and uses them in control [11–
13]. The example of devices with this technology are hybrid bionic systems described
in [14]. This solution may also be combined with a fuzzy controller [15].
The hand prosthesis may be also used by the robot or other similar device in the
teleoperation. Such solution is proposed in this paper, which allows to manipulate
prosthesis simultaneously using special control glove with sensors. The user can
control the artificial hand using his own hand. In addition an tactile feedback is
implemented, using the force sensing resistors located in the artificial hand. In the
design the vibration motors placed in control glove are applied. The design was made,
regarding knowledge provided in the papers [16–19].
2 Kinematic Model
The movement of artificial hand finger is provided by a combination of double four-bar

linkage mechanism (Fig. 2) and has been studied in accordance with [20]. The finger’s
deflection depends on following angles: h1 ; h2 ; h3 and for second part: h01 ; h02 ; h03 ,
whose trajectory is determined by the length of the pushers (Fig. 2 marked as p1 and
p2). The values of c1 and c2 are respectively equal to: 118 and 177 .
K1 cosh3 K2 cosh1 þ K3 ¼ cosðh1 h3 Þ ð1Þ
K4 cosh1 K1 cosh2 þ K5 ¼ cosðh1 h2 Þ ð2Þ
where:
d
K1 ¼ ; ð3Þ
a
d
K2 ¼ ; ð4Þ
c
a2 b2 þ c 2 þ d 2
K3 ¼ ; ð5Þ
2ac
d a2 b2 þ c2 d 2
K4 ¼ ; K5 ¼ : ð6Þ
b 2ab
According to Fig. 2 first four-bar linkage mechanism: a ¼ l1 ; b ¼ r; c ¼ p1 ; d ¼

jO1 ; O2 j and a second one: a ¼ r; b ¼ p2 ; c ¼ r; d ¼ l2 . The angles used to describe
the position of the remaining links are determined as follows:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
B þ B2 4AC
h3 ¼ 2arctg ð7Þ
2A
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
E E 2 4DF
h2 ¼ 2arctg ð8Þ
2D
where:
A ¼ cosh1 þ K3 K1 K2 cosh1 ; ð9Þ
B ¼ 2sinh1 ; ð10Þ
C ¼ K1 þ K3 ð1 þ K2 Þcosh1 ; ð11Þ
D ¼ cosh1 þ K5 K1 þ K4 cosh1 ; ð12Þ
E ¼ 2sinh1 ; ð13Þ
F ¼ K1 þ K5 þ ðK4 1Þcosh1 : ð14Þ
The movement of a second four-bar linkage depends on first one so there is a need
to determine h01 angle. It can be counted by using following equation:
h01 ¼ 180 þ c2 h2 h1 c1 ð15Þ
According to previously mentioned procedure, other angles can be estimated in the

same way. Correctness of described here analysis was checked using CAD model in
SolidWorks software with additional Motion plugin. It also allowed to simulate tra-
jectory of each component and define certain angles. Confirmation of analytical results
with computer simulation was the starting point for further analysis of the discussed
mechanism. Subsequent influence of respective components’ dimensions on a move-
ment are considered and analyzed.
Fig. 2. Finger kinematics with the indication of dimensions of prosthesis’ elements (a) and
angles occurring in an upright position (b).
3 Construction of the Mechanical Part
The design of the artificial hand was prepared in CAD software. The drawings are used
to make a prototype using Fused Deposition Modeling (FDM) method on a 3D-printer.
The elements are made from Polylactide (PLA). The artificial hand consists of a base,
four fingers and a thumb (Fig. 3). The fingers are driven by servo motors. The thumb
has two drives, which ensure the rotation in the plane of the hand and the deflection.
The micro servos were used as the source of the drive due to small dimensions and
small mass. Furthermore, axes are equipped with small bearings to provide low friction
and smoother movement of each component.
The finger consists of three basic elements which geometry and the number of
rotation axes coincide with the human equivalent hand (see Fig. 2 marked as l1, l2, l3).
Dimensions of individual components are equal to: l1 = 48 mm, l2 = 40 mm,
l3 = 34 mm. In addition, two pushers can be distinguished in the design: the first one
connects middle element with the base (Fig. 2 marked as p1) and the second one
connects the first finger element with the third (Fig. 2 marked as p2). Such mechanism
allows to obtain the fold of the finger in the case of its deflection, putting only the first
element of the finger in motion. Thanks to this the appropriate trajectory of displace-
ment is obtained.
In case of the thumb, two degrees of freedom are provided. In the real gripping, the
human hand uses the deflection of the four primary fingers to capture the object. The
grip is supplied by a thumb which stops the object displacement. Additionally, the
thumb allows to block the rotation of the objects in the palm of the hand, which
consists of four primary fingers. The construction of the thumb also consists of three
elements responsible for creating a stable grip and two additional pushers. The artificial
hand is equipped with an overlay that connects the thumb-deflecting part with the
rotating one.
The artificial hand base is enabling the rigid attachment of all its components. Its
design was based on the dimensions of the average male hand (216 178 mm when
unfolded). The vertical arrangement of the servos allows additional rigid support of the
structure by screwing them to the base in the lower part. In addition, servos axes are
parallel to the base plane. The part on which the thumb is located has been tilted at an
angle of 45° (Figs. 3 and 4) to ensure a deeper access of the thumb in the palm of the
hand, which reflects to its natural position.
Fig. 3. Structure of the hand prosthesis.

4 Control Glove with a Force Feedback
The force feedback system is implemented in the designed system: artificial hand-
microcontroller-control glove, basing on the information taken from the force sensing
resistors (FSR) placed on fingertips of the hand fingers (Figs. 4a and 5) and by
application of vibration motors used in a control glove (Fig. 5a). Any contact of the
finger with an object causes the generation of adequate proportional to the applied force
signal, which is send to the microcontroller. According to this signal the microcon-
troller generates output signal with Pulse Width Modulation (PWM) which is used to
control of adequate vibration motor in a control glove. The motors speed corresponds
to the force measured by sensing resistors (details in chapter 5). By this way, the more
force is being applied to the artificial hand finger, the faster the motor vibrates. This
vibration signal is recognized by the user, allowing him to feel the caught of the object
and to touch the object with controllable forces in all fingers. In addition, there has been
used a single microswitch as a second source of tactile for a thumb (Fig. 5b). It is
treated as an example of extra bistable touch provider which gives a signal in case of
object appearance.
To control described here artificial hand, a special dedicated glove is designed and
built. Flex sensors were stitched to the external side of each glove’s finger (Fig. 5). To
provide a steering signal for a thumb in hand’s plane, one of the sensors is placed in
internal side of the palm (Fig. 5a). The vibration motors are placed on phalanges of
glove’s fingers (Fig. 5a). By bending of the fingers in a glove a control signal is
generated to the microcontroller, which converts this signal to the artificial hand servos.
a) b) Servo motors
FSRs
Microswitch
Fig. 4. Construction of the prosthesis. (a) - Internal side of the hand prosthesis (FSR placement),
(b) - Outer side of the hand prosthesis (Servo motors and microswitch placement).
a) Flex sensor b) Flex sensors
VibraƟon
motors
Fig. 5. (a) - External side of the glove (vibration motors and flex sensor placement), (b) -
Internal side of the glove (Flex sensors placement).
By this way the hand fingers follow the control glove fingers movements. The
measured control signals are filtered in order to reduce noises, which are caused by
resistance of flex sensors. Data read by the ADC fluctuate constantly in the range from
1.3 V to 4.5 V according to movement of a digit (respectively: straightening and
deflection). This allows the servo to rotate about 20 or 35° (bending the thumb about
90°) thus value for each digit is set individually. The easiest way to deal with this
problem (filter) is to calculate a mean average for a few samples (five or more) of
measured resistance and then take this result to the control. The control system was
implemented on 8-bit microcontroller.
5 Experimental Tests
As can be seen from the kinematic model, the values of angles and the trajectory of the
finger movement depend on the length of its individual elements (chapter 3).
The last phase of the bended finger trajectory depends on the pushers’ dimensions.
It can be smoother when fingertip faces internal side of a palm as purple trajectory
marked in the Fig. 6a. It allows to achieve better performance in compare to situation
when fingertip approaches with its apex on it (Fig. 6a marked red). This can be realized
by reducing length of pusher p2. The main reason of this case is to provide contact of
the force sensing resistors with manipulated object during bending movement of a
digit. The dimension p1 has an impact during the first phase of the movement (see
Fig. 6b) since it is a part of the first four-bar linkage mechanism. According to the
research a greater value of p1 and p2 cause shorter trajectory which is undesirable
(Fig. 6a marked red).
The measured changes of the h1 and h01 angles during movement from upright to
bent position are shown in Fig. 7. The angle of h1 changes to a lesser extent compared
to h01 . What’s more, the angle h01 reaches zero and increases again (the movement
direction changes). This is due to the shift of point C (Fig. 2) above item l2 (Fig. 6b).
Fig. 6. (a) - Trajectory of the index finger depending on pushers – trimmed and smooth, (b) - a
way of determining the trajectory – with indications according to Fig. 2.
The grasping of the object is presented in the Fig. 9 and in Fig. 8 the courses of
force signals obtained from sensors are show. Phases shown in photos in Fig. 9 a, b, c
are the movement of fingers without the object in the hand’s space. In the phases d, e,
and f (Fig. 9) the artificial hand captured an object, during which the measured forces
are changed which is visible in Fig. 8. As can be seen, fingers wrapped the sponge and
their phalanges rest on the surface of the object (Fig. 9) in accordance with results
explained in Fig. 6. The existing conditions create a convenient adhesion of the surface
of the sensors to the object subjected to be pressured. Thumb has the greatest pressure
due to a force provided by servo overlay which ensures direct force emphasise. The
value of the registered data varies from 0 V (in the absence of pressure) to approxi-
mately 3.4 V (at the maximum pressure recorded). This range has been specially
limited due to fluctuations in the sensor indications.
Such conditions contribute to a significant control of the holding object. The pre-
sented course shows the actual situation of object manipulation, with attention to the
transfer of pressure from the fingers. For this reason, the courses are irregular. During
gripping, the force transmitted by the fingers is subject to continuous adjustment and is
strongly associated with the object. Especially because of its structure, but also its
fragility. The nervous system is the center of stimuli determining the strength of the
pressure exerted. In the case of the presented prosthesis, they constitute vibrating
motors that allow the quantitative assessment of the applied pressure. The applied
pressure changes in the time domain even when the object is held with the same force.
It results from its separation into individual fingers. Human fingers do not have the
same grip and strength. It is different in the case of a designed prosthesis, where each
finger has comparable dimensions and is driven by the same servo motor. The pressure
cycles in the case of the prosthesis therefore differ in relation to the ability of the human
hand. For this reason, the values for a ring and a pinky finger are understated.
Fig. 7. The characteristics of the h1 and h01 angles in the field of time - from upright to bent
position.
thumb index finger middle finger

ring finger pinky finger
500
400
ADC units
300
200
100
0
0 1 2 3 4 5 6
t [s]
Fig. 8. The course of the analogue signals over time while applying pressure to the force
sensing resistors with grasping phases according to Fig. 9.
Fig. 9. Phases of grasping the object of manipulation - sponge as an example (d), (e), (f). The
movement of fingers without the object of manipulation in the hand’s space – (a), (b), (c).
6 Conclusion
The article described construction of the hand prosthesis equipped with force feedback
sensors. The authors focused mainly on the device’s kinematics. In construction it was
applied the double four-bar linkage mechanism detailed described in chapter 2. The
parameters have been chosen to obtain the best trajectory of the fingertips (determined
by the dimensions of the individual mechanism links) – chapter 3. An analytical and
numerical analysis allowed to show dependence of the values of angles and dimensions
in relation to one of the most important aspects of the hand’s prosthesis which is its
finger’s trajectory - chapter 5.
Based on the literature review, it can be concluded that the kinematics of this type
of devices is insufficiently described, especially for low-cost constructions, with a
limited number of drives. The authors meet these criteria by presenting these solutions
and results of activities regarding compromise of the usability and price.
The results of research are the starting point for further investigation on the con-
struction of a comprehensive human hand both shedding new light on other ways to
solve finger motion kinematics.
References
1. Zuniga, J., Katsavelis, D., Peck, J., Stollberg, J., Petrykowski, M., Carson, A., Fernandez,
C.: Cyborg beast: a low-cost 3D-printed prosthetic hand for children with upper-limb
differences. BMC Res. Notes 8(1), 10 (2015)
2. Tutak, J., Klatka, B.: Bionic hand prosthesis (in Polish). Acta Bio-Optica et Informatica
Medica Biomed. Eng. 22(1), 24–30 (2016)
3. Musiolik, A., John, A.: Kinematic analysis of the fingers (in Polish). Eng. Model. 9(40),
111–116 (2010)
4. Jacobsen, S.C., Iversen, E.K., Knutti, D.F., Johnson, R.T., Biggers, K.B.: Design of the
Utah/M.I.T. dexterous hand. In: Proceedings of IEEE International Conference on Robotics
and Automation, pp. 1520–1532 (1986)
5. McMinn, R.M.H., Pegington, J., Hutchings, R.T., Abrahams, P.H.: Atlas of human anatomy
(in Polish). (Slovart/Solis) (1994)
6. Leong, D.P., et al.: Prognostic value of grip strength, findings from the Prospective Urban
Rural Epidemiology (PURE) study. Lancet 386, 266–273 (2015)
7. Pons, J.L., Rocon, E., Ceres, R., Reynaerts, D., Saro, B., Levin, S., Van Moorleghem, W.:
The MANUS-HAND dextrous robotics upper limb prosthesis: mechanical and manipulation
aspects. Auton. Robots 16, 143–163 (2014)
8. Dhillon, G.S., Horch, K.W.: Direct neural sensory feedback and control of a prosthetic arm.
IEEE Trans. Neural Syst. Rehabil. Eng. 13(4), 468–472 (2005)
9. Kawasaki, H., Komatsu, T., Uchiyama, K.: Dexterous anthropomorphic robot hand with
distributed tactile sensor: Gifu hand II. IEEE/ASME Trans. Mechatron. 7(3), 296–303 (2002)
10. Mouri, T., Kawasaki, H., Yoshikawa, K., Taka, J., Ito, S.: Anthropomorphic robot hand:
Gifu hand III. In: ICCAS, pp. 1288–1293 (2002)
11. Świtoński, E., Głowacka-Kwiecień, A., Jochamczyk, K., Jureczko, P., Łosień, T.:
Measurement of muscle action potentials in children using the EMG method (in Polish).
Eng. Model. 7(38), 237–242 (2009)
12. Jamal, M.Z.: Computational Intelligence in Electromyography Analysis - A Perspective on
Current Applications and Future Challenges. (InTech), Croatia (2012)
13. Peleg, D., Braiman, E., Yom-Tov, E., Inbar, G.F.: Classification of finger activation for use
in a robotic prostheses arm. IEEE Trans. Biomed. Eng. IEEE TNSRE 10(4), 290–293 (2002)
14. Navarro, X., Krueger, T.B., Lago, N., Micera, S., Stieglitz, T., Dario, P.: A critical review of
interfaces with the peripheral nervous system for the control of neuroprostheses and hybrid
bionic systems. J. Peripher. Nerv. Syst. 10, 229–258 (2005)
15. Vorotnikov, S.A., Strunin, V.S., Vybornov, N.A.: Using EMG patterns for human gait cycle
recognition. Упpaвлeниe и Bыcoкиe Texнoлoгии 3, 147–162 (2013)
16. Rueckert, E., Lioutikov, R., Calandra, R., Schmidt, M., Beckerle, P., Peters, J.: A low-cost
sensor glove with vibrotactile feedback and multiple finger joint and hand motion sensing for
human-robot interaction. In: Proceedings of the IEEE International Symposium on Robot
and Human Interactive Communication (RO-MAN), pp. 99–104 (2016)
17. Prattichizzo, D., Chinello, F., Pacchierotti, C., Malvezzi, M.: Towards wearability in fingertip
haptics: a 3-DoF wearable device for cutaneous force feedback, pp. 506–516. IEEE (2013)
18. Sziebig, G., Solvang, B., Kiss, C., Korondi, P.: Vibro-tactile feedback for VR systems,
pp. 406–410. IEEE (2009)
19. Galambos, P.: Vibrotactile feedback for haptics and telemanipulation: survey. Concept
Exp. Acta Polytechnica Hungarica 9, 41–65 (2012)
20. Web Site. http://etacar.put.poznan.pl/jacek.buskiewicz//dydaktyka/TM/P_TM_WyklMetody
Analityczne.pdf. Accessed 08 Jan 2018
Author Index
A G
Adamczak, Michał, 72 Gąbka, Joanna, 25
Aliev, Khurshid, 248 Gładysz, Bartłomiej, 274
Antonelli, Dario, 248 Gonçalves, Ricardo M. P., 38
Antosz, Katarzyna, 332 Górski, Filip, 104, 127, 137
Grajewski, Damian, 127
B Gusak, Oleksandr, 165
Balog, Michal, 390
Barros, Ana C., 203 H
Besnea, Daniel, 307 Hadaś, Łukasz, 72
Betta, Jan, 190 Holubek, Radovan, 223
Białek, Marcin, 444 Hošovský, Alexander, 287
Buń, Paweł, 104, 137, 342
I
Iakovets, Angelina, 390
C
Ivanov, Vitalii, 165
Cakmakci, Mehmet, 176
Izdebski, Krzysztof, 381
Chlebus, Tomasz, 190
Cofta, Szczepan, 432
J
Constantin, Victor, 307
Jędrzejczyk, Waldemar, 14
Cyplik, Piotr, 72
Jenek, Tomasz, 368
D K
Domański, Roman, 72 Kapłon, Tomasz, 236
Dostatni, Ewa, 422 Kim, Sumin, 381
Dudek, Adam, 119 Kłos, Sławomir, 119, 261
Kluczek, Aldona, 274
E Kluz, Rafał, 332
Englert, Anna, 368 König, Peter, 381
Erickson, G. Scott, 3 Kopowski, Jakub, 298
Kostal, Peter, 62
F Krawczyk, Marta B., 150
Ferduła, Renata, 432 Królikowski, Marcin A., 150
Flotyński, Jakub, 368 Kubacki, Arkadiusz, 213

https://doi.org/10.1007/978-3-030-18715-6
458 Author Index
Kuchta, Dorota, 190 Rizescu, Dana, 307

Kuric, Ivan, 165 Rojek, Izabela, 298, 422
Rothberg, Helen N., 3
L Rumiński, Dariusz, 368
Lazorík, Peter, 287 Ružarovský, Roman, 223
Lecky, Simon, 62 Rybarczyk, Dominik, 444
Liagkou, Vasiliki, 352
Liaposhchenko, Oleksandr, 165 S
Lucas Soares, António, 203 Saków, Mateusz, 406
Schlueter, Nadine, 92
M Simões, Ana C., 203
Machado, Jose, 38 Skomra, Agnieszka, 190
Macko, Marek, 298 Sobrino, Daynier R. D., 62, 223
Madureira, Ana M., 38 Sokhatska, Hanna, 390
Maik, Mikołaj, 368 Stachowiak, Agnieszka, 72
Miądlicki, Karol, 406 Starzyńska, Beata, 137
Michal, David, 62 Strugała, Dominik, 368
Mikołajewski, Dariusz, 298 Stylios, Chrysostomos, 352
Milecki, Andrzej, 213, 236, 444
Moraru, Edgar, 307
T
Mudriková, Andrea, 62
Torres-Guerrero, Francisco, 319
Torres-Treviño, Luis, 319
N
Trojanowska, Justyna, 342
Neira-Tovar, Leticia, 319
Nowak, Patryk, 444
V
P Varela, Maria L. R., 38
Pandilov, Zoran, 127
Patalas-Maliszewska, Justyna, 92, 119, 261 W
Pavlenko, Ivan, 165, 287 Walczak, Krzysztof, 368
Pezer, Danijela, 48 Walczak, Tomasz, 432
Piteľ, Ján, 287 Wichniarek, Radosław, 104
Putnik, Goran D., 38
Z
R Zawadzki, Przemysław, 104
Rewers, Paulina, 342 Židek, Kamil, 287
Rizescu, Ciprian I., 307 Żukowska, Magdalena, 137

Advances in Manufacturing II: Justyna Trojanowska Olaf Ciszak José Mendes Machado Ivan Pavlenko Editors

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Advances in Manufacturing II: Justyna Trojanowska Olaf Ciszak José Mendes Machado Ivan Pavlenko Editors

Hochgeladen von

Copyright:

Verfügbare Formate

Lecture Notes in Mechanical Engineering

More information about this series at http://www.springer.com/series/11236

José Mendes Machado Ivan Pavlenko

José Mendes Machado Ivan Pavlenko

ISSN 2195-4356 ISSN 2195-4364 (electronic)

This volume of Lecture Notes in Mechanical Engineering contains selected

May 2019 Justyna Trojanowska

Adam Hamrol Poznan University of Technology, Poland

Olaf Ciszak Poznan University of Technology, Poland

Stanisław Adamczak, Poland Jens J. Dahlgaard, Sweden

Marta Grabowska, Poland Alejandro Pereira Dominguez, Spain

Leonilde Varela University of Minho, Portugal

Intelligent Manufacturing Systems

Ivan Pavlenko Sumy State University, Ukraine

Tooling and Fixtures: Design, Optimization, Veriﬁcation

Vitalii Ivanov Sumy State University, Ukraine

Advanced Manufacturing Technologies

Jozef Jurko TU Košice, Slovak Republic

The Changing Face of Production Engineering and Management

Special Session Organizing Committee

Damjan Maletič University of Maribor, Faculty of Organizational

Enabling Tools and Education for Industry 4.0

Dorota Stadnicka Politechnika Rzeszowska, Poland

Staff for the Industry of the Future

Magdalena Wyrwicka Poznan University of Technology, Poland

Advances in Manufacturing, Properties, and Surface Integrity

Special Session Organizing Committee

Szymon Wojciechowski Poznan University of Technology, Poland

Special Session Organizing Committee

Monika Dobrzyńska-Mizera Poznan University of Technology, Poland

Advanced Mechanics of Systems, Materials and Structures

Hubert Jopek Institute of Applied Mechanics, Poznan

Virtual and Augmented Reality in Manufacturing

Filip Górski Poznan University of Technology, Poland

Industry 4.0 Concept

Knowledge Management in Open Industrial Virtual

Smart Industry Solutions

Simulation Modeling of Assembly Processes

Artiﬁcial Neural Network-Supported Selection of Materials

G. Scott Erickson1(&) and Helen N. Rothberg2

Abstract. Paper analyzes prospects for digitization in select manufacturing

Keywords: Digitization Big data Knowledge management Intelligence

© Springer Nature Switzerland AG 2019

to teach others. If it can be done at all, it would be in one-on-one circumstances such as

Table 1. Intangible metrics for select heavy industries

Faculty of Management, Czestochowa University of Technology,

Keywords: Human Organization Industry 4.0 Relation Change

Civilizational breakthroughs, globalization and the expansion of opportunities to

© Springer Nature Switzerland AG 2019

countries, e.g.: Nouvelle France Industrielle – in France, Produktion 2030 – in Sweden,

2 The Prospect of the Fourth Industrial Revolution

The following transformations such as: “mass” digitization of processes, decision-

3 Changes in Human-Organization Relation in Enterprises

3.1 Change in the Function of the Employee in the Enterprise

3.2 Competence Needs of Enterprises

Fig. 1. Competence suitability assessment – aggregated results obtained in individual countries.

Fig. 2. The level of possessed competences – aggregated results obtained in individual

Therefore, it is not easy to ﬁnd proper employees anymore. According to the

Department of Laser Technologies, Automation and Production Engineering,

Keywords: Edge computing Industry 4.0

© Springer Nature Switzerland AG 2019

Manufacturing, Cloud computing; 2. Horizontal integration in collaborative networks

2 Cloud Computing vs. Edge Computing

vij ¼ xij þ Uij ðxij xkj Þ; ð6Þ

xij ¼ dj þ rand½0; 1 ðgj dj Þ; ð8Þ

sij ¼ ð1 qÞsij ; 8i ; j 2 ½1; n ð10Þ

sip ðiÞ ¼ sip ðiÞ þ D; 8i 2 ½1; n ð11Þ