Tackling the Test Challenges of

Next Generation ADAS Vehicle Architectures

David A. Hall

Dr. Abhay Samant

Principal Marketing Manager

Section Manager

ni.com

The Tesla Team

All Tesla Cars Being Produced Now Have

Full Self-Driving Hardware
October 19, 2016

Major ADAS Sensor Types

ULTRASONIC

CAMERA

LIDAR

SHORT RANGE RADAR

LONG RANGE RADAR

ELECTRONICALLY
SCANNING RADAR (ESR)

COLLISION
MITIGATION SYSTEM (CMS)

RACAM
(RADAR + CAMERA)

INTELLIGENT FORWARD
VIEW CAMERA (IFV-100)
SOURCE: www.delphi.com

ADAS Sensor Fusion Evolution Delphi Example

ni.com

ADAS Architectures Continue to Evolve

SMART SENSORS/DECENTRALIZED PROCESSING

RAW SENSOR DATA/CENTRALIZED PROCESSING

Sensor
Electronic Control
Module (ECM)

HYBRID SENSOR/PROCESSING
Source: electronics-eetimes

Deep Learning For Self-Driving Cars

Environmental perception is key to
autonomous driving, e.g. lane position
Traditional feature recognition and
image processing techniques dont scale
to needed complexity
Deep neural networks learn efficient
feature representation
DEEP NEURAL NETWORK

Inductive learning leads to evolving software

operation that is challenging to test

The Connected Car

V2V

V2V

Evolution of 3GPP Standards for Mobile Communications

Historically mobile technology has evolved
primarily to deliver higher data rates.
You
Are
Here

However, recent 3GPP standards

evolution has focused more
heavily on connected devices.

Mobile for

LTE V2X

V2X

5G V2X

Mobile for

LTE-M

IoT

EC-GSM
NB-IoT

3GPPs V2X standards in Release

14 and beyond will utilize LTE &
5G radios for V2V communication.
The Mobile IoT standards deliver
lower radio and operating cost for
services like OnStar and eCall.

Evolution of Automotive Radar Technology

24 GHz Radar

77 GHz Radar

79 GHz Radar

(shared spectrum)

(dedicated spectrum)

(dedicated spectrum)

200 MHz Bandwidth

1.5 m Resolution

600 MHz Bandwidth

0.5 m Resolution

4 GHz Bandwidth
0.1 m Resolution

1971

1993

2004

First blind-spot detection (BSD) radar

at 16 GHz presented at Bendix Labs
in Southfield, Michigan

Radar-based 24 GHz collision

warning system reduces accidents
by 21% for Greyhound busses

European Commission
harmonized use of 79 GHz
band for short-range radar

2006
Field trials of 60 GHz
automotive radar systems on
Jo-Shin-Etsu Highway in Japan

2016
National Highway Traffic Safety Administration
announces that all automakers will supply
collision avoidance by 2022

1959

1975

1998

2006

2014

General Motors integrated a modified

airplane radar to indicate braking
distance on Cadillac Cyclone

First road testing vehicles

Using 35 GHz radar for automatic
cruise control and anti-collision

Mercedes-Benz DISTRONIC
System uses 77 GHz band

Mercedes-Benz DISTRONIC PLUS uses

24 GHz radar for short range 77 GHz
band for long range

Delphi ships 1 Millionth 77 GHz

electronically scanning radar (ESR)

Testing ADAS and the Autonomous Vehicle

MODULE

CAR

CONNECTED CAR WORLD

Characterization
Validation
Software (HIL)
Track and Road
Production

Requirements For An Autonomous Vehicle Testing Platform

Wide range of I/O DC to RF, Vehicle Networks, Wireless Communications

Highly adaptable software to meet changing requirements and architectures

Real time performance to enable emulation of real-world conditions and

HIL testing

System-wide timing and synchronization to correlate across full range of

I/O and communications

Connectivity to standard design, simulation and scenario generation tools

Approaches for Test and Measurement

CLOSED
APPROACH
Vendor knows best

PLATFORM
APPROACH
Customer knows best

Fixed-functionality box
instruments

Out of box measurement

ready

Extend with vendor

accessible customization

Extend with user

accessible customization

Fixed software applications

Productive, approachable
software tools

Monolithic instrument
design

Modular instrument design

with consistent APIs

Resulting closed vendor

ecosystem

Resulting open partner,

user, and IP ecosystem

Software is the Key to Customization

1993

1986
Release on
Macintosh

Compiled
language

1990

Macintosh ,
Sun & Windows

1997

FPGA
programming

Real-Time
enabled

Software
engineering

2006

2003

1998

Internet
ready

2000

Distributed
system
design

2005

Textual math
object-oriented
programming

Multicore
Optimized

2007

Web services

Focus on
software
performance

Models of
computation

2009

2014

2012

2010

2008

Focus on
stability for
long-term
systems

2011

Proficiency
integration
including eLearning

Data-driven
decision
capabilities

Productivity
features focus

2013

Multi-rate Diagram &

High Level Synthesis

2015

20 Man-Millennia of Test and Measurement Software Expertise

Flexibility of PXI for Automotive Design and Test

V2X Testing
GNSS Simulation
Infotainment Test
Radar Target Simulation
Hardware-in-the-Loop
Bus Interfacing

Introduction to Radar Target Simulation

ni.com

Principles of FMCW Radar

frequency

STATIONARY SINGLE TARGET

time

range R
fbu

fbd

fbu

fbd

Beat frequency components due to range

and Doppler frequency shift.

Principles of FMCW Radar

frequency

MOVING SINGLE TARGET

radial velocity vr

f fD

time

range R
fbu

fbd

fbu

fbd

Beat frequency components due to range

and Doppler frequency shift.

PASSIVE TARGET SIMULATOR

ACTIVE TARGET SIMULATOR

Rx

Tx

RF
Control Unit

Delay
Line

Radar System

Radar System
Target Simulator

Simple set-up
Best for fixed targets
Simulates fixed range of distances
Limited doppler and radar cross section

Target Simulator

Simulate multiple independent targets

Advanced Doppler simulation
Variable target control
Complex scenarios possible
Minimum distance dependent on
RF control unit delay

Block Diagram of Active Radar Target Simulator

ADC

PXI System
Upconverter and
Downconverter for
Frequency Extension

RFIN

LabVIEW Programmable FPGA

0
90

ADC

Controller
DMM

NI 76-81 GHz
ADAS Radar
System for
Automotive

DAC
VST

RFOUT
0
90

DAC
PXI
Programmable
Delay Module

Digital Interface

NI Vector Signal Transceiver (VST) simulates targets

using LabVIEW FPGA-based signal processing
Doppler shift via Tx-to-Rx frequency offset
Distance to target via delay
Multiple targets

NI ADAS Test Solution

Features at a Glance

Simulates up to 4 targets and settings include

velocity, RCS, and angle of arrival

Measurements include: radiation pattern, EIRP,

phase noise, spectrum occupancy, beam width,
and chirp analysis

Flexible software-based approach allows users

to create custom target scenarios

Key Specifications

Frequency Range: 76 81 GHz

Target range: 1 to 250 meters

Range resolution: 0.1m

Velocity: 0 to 200 km/hr

Bandwidth: 2 GHz for short-range, 1 GHz for long-range

The NI ADAS Test Solution is based on

NIs 2nd generation Vector Signal Transceiver and
uses a LabVIEW-programmable FPGA for target emulation.

NEW!

2nd Generation VST: PXIe-5840

Specification

PXIe-5840 Performance

Frequency Range

9 kHz to 6.5 GHz

Max. Output Power

+20 dBm

Bandwidth

1 GHz

EVM

-50 dB
(802.11ax, Loopback, external LO)

Tx/Rx Amp. Accuracy

0.5 dB / 0.5 dB

Tuning Time

300 us

Slots

FPGA

VIRTEX 7 690T

Digital I/O

60 MHz, 8 port high-speed parallel

12 Gbps, 4 port high-speed serial

Programmable FPGA allows you to execute

LabVIEW code onboard the instrument

A Single Platform for RF and HIL Test

CAN, FlexRay, and Ethernet

Radar

ADAS
ECU

Real-Time Test (HIL) Software

V2X
Module

Real-time OS
Mathematical Modeling
Scenario Simulation
Camera

COPYRIGHT SAAB AB

With the PXI Vector Signal Transceiver, the combination of the industrys widest bandwidth and low
latency software has allowed us to discover our automotive radar sensors as never before.
Niels Koch, Audi

ni.com

Konrad
77 GHz Radar Target Simulator

System Features

Scenario Editor Software

Konrad ABex Production System uses

National Instruments ADAS Test System
19 6HE Rack, all electronics and RF
components integrated in one box.
Antenna blind mate connectors for remote
antenna connection.

Supports creation of custom scenario

generation and tracks
Sequence Tracks to simulate various radar
environments

ni.com

Typical Target Simulation Scenarios

S TA R
PAT T E R N

LANE
CHANGE

OBJECT
CROSSING STREET

Approaches for Test and Measurement

CLOSED
APPROACH
Vendor knows best

PLATFORM
APPROACH
Customer knows best

Fixed-functionality box
instruments

Out of box measurement

ready

Extend with vendor

accessible customization

Extend with user

accessible customization

Fixed software applications

Productive, approachable
software tools

Monolithic instrument
design

Modular instrument design

with consistent APIs

Resulting closed vendor

ecosystem

Resulting open partner,

user, and IP ecosystem

Summary

Autonomous driving is quickly becoming a reality

Technologies, architectures and requirements are evolving rapidly

RF and communications technologies and standards are playing a central role

Complexity of systems and applications requires a platform - based approach to test