Available online at www.sciencedirect.

com

Procedia Engineering 48 (2012) 592 – 598

MMaMS 2012

Advanced control of the mono axial vehicle with variable parameters

Jozef Rodinaa*, Peter Hubinskya
a
Institute of control and industrial informatics, Faculty of Electrical Engineering and Information Technology, Slovak University of Technology, Ilkoviþova
3, Bratislava, 812 19

Abstract

The Mono Axial Vehicle (MAV) mobile robot platform is two wheel chassis with ability to change of the position of the center of gravity.
Our mathematical model brings a new approach to modelling of these types of vehicles. We developed one generalized mathematical
model with generalized distance of the center of mass of the MAV. Thanks to this approach we are able to make controller synthesis
much easier than before. We also deal with robustness of this controller and observer for the generalized distance of the center of mass of
the MAV.
© 2012 The Authors. Published by Elsevier Ltd.
© 2012 Published by Elsevier Ltd.Selection and/or peer-review under responsibility of the Branch Office of Slovak Metallurgical Society at
Selection
Faculty and/or peer-review
of Metallurgy under
and Faculty responsibility
of Mechanical of the Branch
Engineering, Office
Technical of Slovak
University Metallurgical
of Košice Society
Open access underat
CCFaculty of Metallurgy
BY-NC-ND license. and
Faculty of Mechanical Engineering, Technical University of Košice.

Keywords: Mono Axial Vehicle , MAV, inverted pendulum, dynamical model, MEMS

Nomenclature
Mp mass of the wheels (kg)
mp mass of the body (kg)
Jw inertia of wheels (kg.m2)
R radius of the wheels (m)
L distance of the center of mass from the centre of the body
ș angle of deviation of the body (rad) also derivations of this angle are used
ij angular position of the wheel (rad) also derivations of this position are used

1. Introduction

In this paper we present the Mono Axial Vehicle (Fig. 1.) as mobile robot platform. Differences between this mobile
robot platform and well known personal vehicle Segway™, which actually also have just two wheels, is that AMAV
platform has ability to change the position of center of the gravity over the wheel axis [1]. Use of the MAV in recon type of
mobile robots needs some recon equipments consist of Infrared camera basic camera system. We also plan to use Kinect™
sensor for reactive navigation [2] of the robot. Recon equipments will be mounted on retractable platform (Fig 2.) which can
be hidden in to the body of the MAV or can move out from the body of the robot. During movement of this retractable
platform center of the gravity of the body is moving too. Therefore this system is changing from stable system to unstable

* Corresponding author.
E-mail address: jozef.rodina@stuba.sk. 

1877-7058 © 2012 Published by Elsevier Ltd.Selection and/or peer-review under responsibility of the Branch Office of Slovak Metallurgical Society at
Faculty of Metallurgy and Faculty of Mechanical Engineering, Technical University of Košice Open access under CC BY-NC-ND license.
doi:10.1016/j.proeng.2012.09.558
 Jozef Rodina and Peter Hubinsky / Procedia Engineering 48 (2012) 592 – 598 593

system positions. Instead of use two dynamical models for stable and unstable position we have develop one generalized
dynamical model of the MAV which will be described in the next chapter.

2. Dynamical model of the MAV

For purpose of modelling we made simplification of the MAV. We will consider only straightforward motion of the
mobile robot. First we define system of the MAV with its parameters according to the figure 3.

Fig. 1. 3D model of the mono axial vehicle (MAV).

CG

CG

Fig. 2. Retractable platform for recon equipments on the MAV.

From this system definition is obvious that system is described like pendulum on the wheel. In the wheel – pendulum
joint is propulsion system (DC motor).
y

180°
mw Jw

R
x
0
L
ș mp

0°

Fig. 3. Parameters and coordinate system of the MAV.

594 Jozef Rodina and Peter Hubinsky / Procedia Engineering 48 (2012) 592 – 598

At this stage of modelling we do not take into account external forces acting on the MAV. In order to create dynamical
model of the MAV we have to use Lagrange equation of the second kind [3] [4]. We need to obtain potential and kinetic
energy from the MAV system.

For wheels we have:

Kinetic energy translational

1
EKT = mWv1
2
(1)
2
2
§ • ·
¨ϕ ¸
v =¨
1
2
¸¸ (2)
¨R
© ¹

Kinetic energy rotational

1 • 2
E KR = JW ϕ (3)
2
Potential energy

EP = 0 (4)

For body we have:

Kinetic energy
1 • 2 • • 1 •2
E K = m P ϕ R + m P LR ϕ cos θ θ + m P L θ
2 2
(5)
2 2
Potential energy

EP = mP gL(1 − cosθ ) (6)

Lagrangian of the system (L = T - U)

1 •2 2 1 • 2 • • 1 •2
L = ϕ R (mW + mP ) + J W ϕ + mP L ϕ R θ cos θ + m p L θ − m p gL + m p gL cos θ
2 (7)
2 2 2
From partial derivation of the Lagrange of the MAV we get equations describing dynamics of the MAV.

d §¨ ∂L ·¸ ∂L
− =0
dt ¨ ∂ ϕ• ¸ ∂ϕ (8)
© ¹

d §¨ ∂L ·¸ ∂L
− =0
dt ¨© ∂ θ• ¸¹ ∂θ
2

(m R + m R + J )ϕ+ m LRθ cosθ − m LRθ

•• •• •

W
2
P
2
W P P sinθ = 0 (9)

•• ••
m P LR ϕ cos θ + m P L2 θ + m P gL sin θ = 0 (10)

When we get into account external torque we get following equation of motion [5].
 Jozef Rodina and Peter Hubinsky / Procedia Engineering 48 (2012) 592 – 598 595

(m R )
•• •• •
W
2
+ mP R2 + JW ϕ+ mP LRθ cosθ − mP LRθ sinθ = Mp (11)

•• ••
m P LR ϕ cos θ + m P L2 θ + m P gL sin θ = − Mp (12)

We do not take into account dumping and frictions acting in the MAV system at this moment, for simplification of control
structure synthesis.

3. Linearization and Control synthesis

3.1. Linearization

For linearization of equations 11 and 12 we put cosș equal to 1, sinș equal to ș, and ș’ is equal to zero. We get following
linear equations:
•• ••
J s ϕ + m P LR θ = Mp (13)
J s = mW R + m P R + J W 2 2

•• ••
m P LR ϕ + m P L2 θ + m P gL θ = − Mp (14)

When we rearrange equations 13 and 14 we get:

••
•• M m p LR θ
ϕ = p
− (15)
Js Js
•• M R •• g
θ =− p
2
− ϕ− θ
mpL L L (16)

From these equations we create locally linearized dynamical model of the MAV in Matlab Simulink as can be seen in the
figure 4.

Fig. 4. Linearized dynamical model of the MAV.

3.2. Controller structure design

Our presented dynamical model is system of fourth order with following state variables:
596 Jozef Rodina and Peter Hubinsky / Procedia Engineering 48 (2012) 592 – 598

ªϕ º
«•»
ϕ
x=« »
(17)
«θ »
«•»
«¬θ »¼
We chose modal control structure design as can be seen in the figure 5. Also major parameters of system such a dynamic
of current controller, gain of the drive, DC motor constant and current limit were added in this control structure.

Fig. 5. MAV controller structure design.

As can be seen from figure 5 there are four proportional controller gains l1 to l4 and one compensation gain –l1 (later in this
text will be explained this compensation gain). We chose pole placement method for gain setup. First we need to put
characteristic polynomial A(s) equal to our desired system parameters as follow.
!
A ( s ) = ( s 2 + 2 b1ω 01 s + ω 01 )( s 2 + 2 b2ω 02 s + ω 02 )
2 2
(18)

We choose system parameters as follow:

1 1 g g 1
b1 = ; b2 = ; ω 02 = ;ω 01 > *
2 2 R R + Lmin ω 02
!
R + Lmin > 0

When we solved equation 18 with our system parameters (masses wheel diameter etc.) we get control gains for the system
(Fig. 5). Solution of the equation 18 depending also on L parameter of the system, which is actually variable parameter. In
real system we can obtain L from look-up table with pre-measured values, or we can make an observer of L from the
appropriate set of state variables. Because discussed system is static system there will be nonzero value in steady state of the
system.
1
Vale of steady state will be equal to
− l1
Therefore we need to use steady state gain compensation which is equal to –l1.
 Jozef Rodina and Peter Hubinsky / Procedia Engineering 48 (2012) 592 – 598 597

4. Simulations

We perform two sets of simulation. First set is with L > 0 and second set is with L < 0. This mean that in case when L > 0,
centre of the gravity is located under the wheel/body axis and thus system is stable. In case when L < 0, centre of the gravity
is located over the wheel axis and system became unstable and is acting like a inverted pendulum on a wheel. In the figure 6
can be seen set of graphs for first case (L > 0) and in the figure 7 can be seen opposite case (L < 0). From these graphs is
evident that our pole placement controller design is suitable for both positions of the centre of the gravity (stable system /
unstable system).
Of course there are some limitations of this controller regarding with parameters of used drive in compare with parameter
of system itself. For example in case when L < 0 and current will saturate there are none possibilities how to regulate this
system and the real MAV will fall down to the ground. Other limitation of this controller is when L = 0. I n this
0.15
1.2
Real position
Desired position 0.1
1

0.05

Body deviation angle [rad]

0.8

0
Position [m]

0.6

0.4 -0.05

0.2 -0.1

0 -0.15

-0.2 -0.2
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
Time [s] Time [s]

1.5

0.5
Current [A]

-0.5

-1

-1.5

-2
0 1 2 3 4 5 6 7 8
Time [s]

Fig. 6. Graph set for simulation with L > 0.

1.2 0.06
Real position
Desired position
1
0.04

0.8
Body deviation angle [rad]

0.02
Position [m]

0.6
0
0.4

-0.02
0.2

-0.04
0

-0.2 -0.06
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
Time [s] Time [s]
598 Jozef Rodina and Peter Hubinsky / Procedia Engineering 48 (2012) 592 – 598

2.5

1.5

0.5

Current [A]
0

-0.5

-1

-1.5

-2

-2.5
0 1 2 3 4 5 6 7
Time [s]

Fig. 7. Graph set for simulation with L < 0.

case system is on stability bound. We can avoid this state by using of hysteresis in surround of zero for controller gain
calculation.

5. Conclusion

In this paper we present a generalized dynamical model of the two wheel vehicle known also as the Mono Axial Vehicle.
We linearized this dynamical model in order to simplification of controller structure synthesis. We chose modal control
structure and design parameters of this controller using pole-placement method. Provided simulations proved that this
controller is able to work in stable and also in unstable position of the center of the gravity of the MAV. In the end we
discus about controller limitations and we propose some solutions to deal with this limitations. Our future work is oriented
on deployment verification of designed controller into real MAV robot.

Acknowledgements

This work was supported by Grant No. APVV-0261-10. The authors are pleased to acknowledge this support.

References
[1] Grasser, F., D’Arrigo, A., Colombi, S., Rufer, A., 2001. JOE: A Mobile, Inverted Pendulum. [online]
http://leiwww.epfl.ch/publications/grasser_darrigo_colombi_rufer_mic_01.pdf.
[2] Babinec, A., Vitko, A., DuchoĖ, F., Dekan, M., 2011. Reactive Navigation of Mobile Robot Using Vector Field Histogram+. In: Modelling of
Mechanical and Mechatronic Systems. MMaMS 2011 [elektronický zdroj] : Proceedings of the 4th International Conference. HerĐany, Slovakia. 20.-
22. Sept. 2011. - Košice : Technická univerzita v Košiciach, 2011. - ISBN 978-80-553-0731-2. - CD-Rom.
[3] Schaub, H., Junkins, J.L., 2003. Analytical mechanics of aerospace systems. Virginia : American Institute of Aeronautics and Astronautics, Octobre
2003. ISBN-13: 978-1563475634.
[4] Claerence, W.D.S., 2005. Vibration and Shock Handbook. Boca Raton : CRC Press, June 2005. ISBN-13: 978-0849315800
[5] Gmiterko, A., Vacková, M., 2011. Dynamic model of vehicle with two coaxial parallel wheels. IEEE 9th International Symposium on Applied
Machine Intelligence and Informatics (SAMI), Smolenice, Slovakia, Jan 27, 2011 - Jan 29, 2011, s.303 – 305, ISBN: 978-1-4244-7429-5