Senses for Safety. - de.mathworks.com filePublic Dr. Alexander Behrens ADAS Business Unit Global...

24

Transcript of Senses for Safety. - de.mathworks.com filePublic Dr. Alexander Behrens ADAS Business Unit Global...

Wenn Sie ein neues Bild einfügen: Klicken Sie mit der rechten Maustaste auf das Bild und wählen „In den Hintergrund“, um das Bild hinter das Quality Seal zu bringen.

If you insert a new picture: Right click on the picture and select “Send to the back” to position it behind the Quality Seal.

Senses for Safety.

Driver assistance systems help save lives.

MATLAB Expo 2014

Verkehrszeichenerkennung in Fahrerassistenzsystemen

– MATLAB @ Continental –

http://www.continental-automotive.com/

Chassis & Safety Division

Advanced Driver Assistance Systems

Public

Dr. Alexander Behrens

Continental Corporation Five Strong Divisions

09 July 2014

3

Chassis &

Safety

Vehicle Dynamics

Hydraulic

Brake Systems

Passive Safety &

Sensorics

Advanced Driver

Assistance Systems

(ADAS)

Powertrain

Engine Systems

Transmission

Hybrid Electric

Vehicle

Sensors &

Actuators

Fuel Supply

Interior

Instrumentation &

Driver HMI

Infotainment &

Connectivity

Body & Security

Commercial Vehicles

& Aftermarket

Tires

PLT, Original Equipment

PLT, Repl. Business, EMEA

PLT, Repl. Business, The Americas

PLT, Repl. Business, APAC

Commercial Vehicle Tires

Two Wheel Tires

ContiTech

Air Spring Systems

Benecke-Kaliko

Group

Compounding

Technology

Conveyor Belt

Group

Elastomer Coatings

Fluid Technology

Power Transmission

Group

Vibration Control PLT – Passenger and Light Truck Tires

Public

Dr. Alexander Behrens

ADAS Business Unit Global Footprint

09 July 2014

4

Seguin, USA

SOP 2014

Ingolstadt, GER

Yokohama, JAP

Timisoara, ROM

Lindau, GER

Headquarters

Lead Plant

Lewes, GBR

Engineering

Production

Sales Ottobrunn, GER

Auburn Hills, USA

Nuernberg, GER Regensburg,GER

Sibiu, ROM

Calamba, PH

SOP 2015

SOP 2016

Karben, GER

SOP 2016

Ulm, GER

Bangalore, IND

Public

Dr. Alexander Behrens

ADAS Functions Overview & Motivation

09 July 2014

5

“70% of all serious

accidents could potentially

be avoided by driver

assistance systems”

(BAST Germany)

Increase safety & comfort is our

mission!

Lane Departure Warning Emergency Brake Assist Adaptive Cruise Control Traffic Sign Assist

Blind Spot Detection Intelligent Headlamp

Control Rear Cross Traffic Alert Surround View

Public

Dr. Alexander Behrens 09 July 2014

6

Multi Functional Camera

Lane Recognition Light Sensing Traffic Sign

Recognition Vehicle

Detection

One Camera for

Multiple Functions

Pedestrian

Detection

Public

Dr. Alexander Behrens 09 July 2014

7

Challenges – Getting confused?

Public

Dr. Alexander Behrens

Challenges – Sign Set

09 July 2014

8

› Recognized sign types with and without navigation input (with examples)

› Speed limits

› No-passing

› Cancellations

› Electronic

› Directional/ no entry

› Stop signs

Public

Dr. Alexander Behrens

Circular Sign Recognition Algorithm

09 July 2014

9

Video Camera

Grayscale Image

Public

Dr. Alexander Behrens

Circular Sign Recognition Algorithm

09 July 2014

10

Video Camera

Grayscale Image

Edge Detector

Public

Dr. Alexander Behrens

Circular Sign Recognition Algorithm

09 July 2014

11

Video Camera

Grayscale Image

Edge Detector

Hough Transformation

Public

Dr. Alexander Behrens

Circular Sign Recognition Algorithm

09 July 2014

12

Video Camera

Grayscale Image

Edge Detector

Tracker

Hough Transformation

Public

Dr. Alexander Behrens

Circular Sign Recognition Algorithm

09 July 2014

13

Video Camera

Grayscale Image

Edge Detector

Tracker

Classification

Hough Transformation

Public

Dr. Alexander Behrens

Circular Sign Recognition Algorithm

09 July 2014

14

Video Camera

Grayscale Image

Edge Detector

Tracker

Classification

Result:

Hough Transformation

Public

Dr. Alexander Behrens

› MATLAB Tools are developed and used for

› Development / training, analysis, review, evaluation

› Interaction with different databases

› Advantages

› Easy and fast pre-development

› Evaluation and generation of key performance indicators

› GUI and user friendly interfaces

› Example: Classifier Training

› Training of classifiers for Traffic Sign Recognition is data driven

› Automated training

› Handling of large databases

› Evaluation and analysis

MATLAB Tooling Example: Traffic Sign Recognition - Classifier Training

09 July 2014

15

Public

Dr. Alexander Behrens 09 July 2014

16

Labeled Data

“Ground Truth”

Recording /

Algorithm

Output

MATLAB

Toolchain

Traffic Sign

Assist

Algorithm

Camera

Label Inspector

Labeling

Scene Inspector

Public

Dr. Alexander Behrens

Scene Inspector

09 July 2014

17

Public

Dr. Alexander Behrens

Scene Inspector

09 July 2014

18

Public

Dr. Alexander Behrens

Scene Inspector

09 July 2014

19

Public

Dr. Alexander Behrens 09 July 2014

20

Labeled Data

“Ground Truth”

Recording /

Algorithm

Output

MATLAB

Toolchain

Traffic Sign

Recognition /

Classification

Traffic Sign

Assist

Algorithm

Scene Inspector

Camera

Label Inspector Sample Inspector

Labeling Synthetic

Sign Data

Public

Dr. Alexander Behrens

Sample Inspector

09 July 2014

21

Public

Dr. Alexander Behrens

Evaluation

09 July 2014

22

MATLAB

Toolchain

- Image Processing Toolbox

- Signal Processing Toolbox

- Statistics Toolbox

- Curve Fitting Toolbox

- Parallel Computing Toolbox

Public

Dr. Alexander Behrens

› MATLAB is used in daily work for development and evaluation of driver assistance

functions

› Prototypes are designed with MATLAB for predevelopment and proof of concept

› Data management, evaluation, and interactive analysis are supported by MATLAB tools

and GUIs

› Traffic Sign Recognition and other functions make high use of MATLAB tools

› MATLAB and its established features

› reduces our tool development efforts,

› accelerates our simulation cost,

› and allows reliable, repeatable and accurate parameter optimizations

› Engineers having good MATLAB programming skills are highly requested

Summary

09 July 2014

23