Online and Offline Validation of ADAS ECUs · 2020. 7. 15. · Online and offline verification by...

V1.0 | 2020-07-10

With CANape Option Driver Assistance

Online and Offline Validation of ADAS ECUs

2

u Introduction

Validation of ADAS ECUs

Time Synchronized Recording

From Signals to Data-Objects

Object Overlay and GFX Configuration

Analyzing of Measurement Files

Capture Data-Objects and Raw-Data from ADAS Sensors

More Information about Vector ADAS Products

Agenda

3

What are Advanced Driver Assistance Systems?

Introduction

u Driver assistance systems are electronic components in vehicles for

u Different sensors acquire the vehicle's surroundings

u The sensor data is then analyzed and merged in the ECUs

u Large quantities of data from different sensors must be visualized and validated

u Radar, lidar, ultrasonic, laser and video-based systems

u Assisting the driver

u Enhancing safety

u Improving convenience and economy

4

CANape Solutions for Advanced Driver Assistance Systems (ADAS)

Introduction

ADAS Road and Laboratory Validation

Online and offline verification by overlaying objects on video, map and scene view

ADAS Logging

Scalable ADAS Logging Software and Hardware

5

Introduction

ADAS ECU road and lab validation

u Validate object recognition algorithms

u Implemented in ECU e.g. ACC, "stop and go" systems, lane detection …

u With the help of object overlaying

u Online during test drives and offline with recorded data

u Typical user groups:

u OEM engineer who receives ADAS system/ECU from supplier > Validation of ECU functionality

> Fine tuning of parameterization for a particular vehicle

u Engineer at supplier with similar assignment > Validation & calibration

> No development of algorithms inside the ADAS system

ADAS Logging

Different Use-Cases

6

Introduction

ADAS ECU road and lab validation

u High-end logging of complete ADAS car sensorics

u Collecting ADAS sensor raw data or ECU internal data

u From different sensor vendors

u Handling of high data rates and time synchronous recording

u Typical user groups:

u OEM engineer who develops autonomous driving cars> Running of test fleets for collecting data

> Validation of ECU functionality in simulations with real measured data

u Engineer at supplier with similar assignment

ADAS Logging

Different Use-Cases

7

Introduction

CANape - Time synchronized measurement and calibration of ECUs

ADAS sensor:

u Radar

u Lidar

u Infrared

u Video

u Ultrasonic

DriverAssistance

ECU

Video/GPS

u Synchronized measurement and recording of

u ECU internal Signals

u Radar, Lidar and other sensors

u Video sources

u GPS devices

u …

Measurement:

u Video camera

u GPS

Vector solution for measurement, calibration and validation

8

u It’s difficult to validate a radar object in a graphical or numerical window

u There are several questions like

u Is this detection correct?

u Is the detected object a car, pedestrian, tree…?

u Where are all detections?

u …

ADAS Validation

Introduction

CANape Option “Driver Assistance”

9


CANape - Time synchronized measurement and calibration of ECUs

+ Option “Driver Assistance”

Vector Solution for Measurement, Calibration and Validation

ADAS sensor:

u Radar

u Lidar

u Infrared

u Video

u Ultrasonic

DriverAssistance

ECU

Video/GPS Measurement:

u Video camera

u GPS

10

Introduction


u Time Synchronized Recording






Agenda

11

Recording of Sensor Data and Video Sources (Context Camera)


u ADAS ECU delivers the results in signals or as data-objects (serialized signal stream) on busses

u Position in X-, Y- and Z-coordinates

u Detected lanes and position

u Signs …

u Additional integrated cameras in the car recording the situation

u CANape measures the ECU data via CAN, Ethernet, XCP on CAN, VX1000 …

u Time synchronization of video frames to other measurement data

u CANape is optimized to handle high frame rates with low CPU usage

CANapeAVI

Optimized frame rate for best CPU performance for visualization

Storing and data handling with original frame rate of camera

12



u Adding camera device in “Device Configuration”

u New camera device

u Multimedia signal is available

u Multiple cameras can be added> USB Cameras with Direct show

> AXIS Cameras with F44 main unit

u Multimedia signal

u can be assigned to recorder

u storage path can be set

13



u Multimedia signal in “Video Window”

u Live view during run-time

u Global measurement cursor

u Support of different compressors

14

Introduction



u From Signals to Data-Objects





Agenda

15

u A data-object is an instance of defined classes

u It could be a measurement data-object or a measurement file data-object

u Existence of data-objects is optional

u Classes

u Definitions of one or multiple data-objects

u Can consist of structures

u Data-object list/container

u List of multiple instances of a class

u Often the data-objects are send

u serialized and signal based

u distributed to many messages

Deserializing of the data-objects on receiver side necessary

Definition of Data-Objects


CAR_2 CAR_1

Obje

ct

„CAR“

Relative Velocity Long

Relative Velocity Lat

Displacement Long

Displacement Lat

Orientation Angle

Size Length

Size Width

16

u Creation of new data-objects and classes based on signals with the “Signal-Object-Adapter”

Object-Oriented with “Signal-Object Adapter”


Buffer

Buffer

Buffer

MDF

Buffer

u GUI-based creation of new classes and Data-Object instances

u One Signal-Object Configuration for each device or MDF4 file

1

2

17

u For the conversion from signals to objects a converter type could be defined

u Depending on serializing type on transceiver site

u There are three ways that data objects can be send on the bus



u Default

u Trigger point (TP) of an object or object list is reached after each signal is received at least once

u Modulation

u Objects are based on modulation of signals

u Count (C) defines for each cycle how many modulations of the signals their have been

18

u For the conversion of objects out of measurement files the converter type could be defined



u Sequence

u Objects based on a sequence of signals

u Count (C) defines for each cycle how many signals for each property are considered

u The last received value of each signal is used

19

u New type for data-objects in the measurement configuration and MDF4 files available

Data-Objects Configuration/Deserializing


Allow the deserialization of data stream into an object based representation

Online during measurement

Offline out of signal based measurement file

20

Introduction




u Object Overlay and GFX Configuration




Agenda

21

u Graphical object overlay for detected ADAS objects like

u vehicles, lane markers, traffic signs … for multiple sensors

u Integration of the new ADAS views which support the data-object-oriented overlay

u Video view, Map view, Scene view and ADAS Explorer

u Object-oriented GFX objects are global in the current project

u No need to define GFX objects for each view

u In the ADAS Explorer the GFX objects could be enabled/disabled for each view

Data-Object-Oriented Overlay


GFX ConfigurationScene View

Map View

ADAS Explorer

22

u Overlay of images / bitmaps to visualize

u Ego- and detected vehicles

u Traffic signs

u User specific icons (danger, attention etc.)

u Available for all views (Scene, Video, Map)

Object-Oriented Object Overlay - Available GFX-Objects


23

GFX Editor to Configure the Graphical Objects


Properties window to change objectproperties

Available Data-Objects in project

List of configured Overlay Objects Type of

Overlay Object and Selection

24

u To calibrate the camera to the test objects, various points with known position must be assigned on the monitor

u Manual or guided video calibration available

u The corresponding coordinates must be specified for each individual point

u Different building place or angles for reference system and original sensor

Video Calibration Process


“Bad” calibrated camera “Good” calibrated camera

25

u New “Vector Camera Calibration Tool”

u Guided calibration process

u Only a chessboard with known dimensions and number of fields is needed

u 10 or more pictures with different angle and distance

u Calibration file is generated for normal and fish-eye cameras

Guided Camera Calibration Process


u Graphical overlay of detected fields and test object for quality check

26

Introduction





u Analyzing of Measurement Files



Agenda

27

Retrieval of Signal Conditions and Automated Analyzing


u More and more measurement files with thousands of signals are recorded

u It is not possible to validate all data live during measurement

u How to find specific conditions and signal combinations in big archives?

u Calculation of functions and scripts based on signals are complex> Not possible to calculate during runtime

> Post calculation to find specific conditions for different use-cases

u Braking lights detected

&Lane Type is solid

u List of all “Hits” u Marker and fast navigation

28



u CANape offers the possibility to load measurement files in the same project used for recording

u Object overlay is possible for online and offline validation

u For video and GPS windows

u Functions and scripts can be used for

u searching for specific condition

u automated analysis

u Printing and reporting of results direct out of CANape

u Data mining functionality for analyzing more than one measurement file

29



u Full support of GFX Editor

u Creation of new objects from signals or function results

u Using possibilities of newest CANape versions with old measurement files

30

Introduction






u Capture Data-Objects and Raw-Data from ADAS Sensors


Agenda

31

The Challenge – Logging of Multi-Sensor and Fusion ECUs of Multiple Suppliers


How to log the complete car environment with data rates of several GBytes/s?

Log 1

Log 3

Log 4

FrontRadar

CornerRadar

FrontCamera

SideLidar

RearCamera

ContextCamera

FusionECU

Time- Trigger- Start/Stop-Synchronisation ??

Log 2 Log 5

u How to handle high data rates of:

u Radar Raw Data

u Laser scanner

u Mono/Stereo cameras

u „Classic“ XCP signals

u Limitations may apply

u Computer resources

u Disk space/writing performance

u CPU usage

u Available sensor integration

32

Time- Trigger- Start/Stop-

Synchronisation √

CANape

Solution: Distributed High Performance Recording

u A special designed standalone recorder for one specific device

u Multiple DHPR can be used in parallel

u One measurement file for each DHPR

u All created MDF files are time synchronous

u Easy access to file distributed file content via loading the complete measurement

u Advantages:

u Integration of new sensors and sources in CANape independent of CANape release

u CANape replaces all individual loggers

u Optimized PC resource and storage usage

u Measurement rates of several GBytes/s

u Synchronized logging distribution to multiple PCs

u Please contact us for project based sensor integration

Solution – CANape as Scalable Recorder for the Complete Environment


33

DHPRs on a Single PC


CANape

• XCP• Buslogging• Reference Camera• etc…

XCPonETH (e.g. Fusion ECU) XCP DHPR

XCP/RIF DHPRVX1000 XCP/RIF (e.g. Radar)

Video DHPRFront camera (ME chip)

Lidar DHPRLidar (UDP decoding)

UDP DHPRCustomer specific (e.g. UDP)

MDF

MDF

MDF

MDF

MDF

MDF RIF

> Time synchronization> Start / Stop Measurement> Trigger> Data backchannelT

CP

All MDF are recorded time synchronous

One MeasurementPC

34

DHPRs distributed on Multiple PCs


• XCP• Buslogging• Reference Camera• etc…

XCPonETH (e.g. Fusion ECU) XCP DHPR

XCP/RIF DHPRVX1000 XCP/RIF (e.g. Radar)

Video DHPRFront camera (ME chip)

Lidar DHPRLidar (UDP decoding)

UDP DHPRCustomer specific (e.g. UDP)

MDF

MDF

MDF

MDF

MDF

MDF RIF

> Time synchronization> Start / Stop Measurement> Trigger> Data backchannelT

CP

All MDF are recorded time synchronous

One MeasurementPC

PC

CANape

35

u Sensors from multiple vendors deliver data via their own Ethernet protocols

u Without using standardized automotive databases/descriptions

u Referring to – and re-using – application software internal data structures

u Highly flexible for the (software) developer with such a data source

→ But: How to bring such data into the measurement tool?

u Solution: CANape Protocol Decoder

u Customizable decoder DLL > Decodes signals and data objects, based on the user specific Ethernet protocol specification

u Plugin in CANape> Plugin = tile in the device configuration

u Supports UDP and TCP

→ Implementation of user-specific Protocol Decoders will be realized by Vector as projects

Support for User-Specific Ethernet Protocols


36

Supported DHPR Devices out of Different Areas


u Radar (RAW Data, XCP, proprietary)

u LIDAR (Scala, IBEO, Quanergy, Velodyne, Hesai …)

u Context/Reference Cameras

u Vehicle Cameras (MobilEye-TAPI, RAW Data …)

u Fusion-ECU / XCP-based Systems

u Analog sensors (pressure, accelerometer …)

u GPS / IMU (GeneSys ADMA …)

u Vehicle Networks

u Other sensors (Brightness, Humidity, Audio)

37

Introduction







u More Information about Vector ADAS Products

Agenda

38

Additional Information about the Vector ADAS Solutions


u Vector offers different solutions to the topic of ADAS development and verification

u Please have a look on our Webpage for products, articles, know-how, webinars and more:

https://www.vector.com/int/en/know-how/technologies/autonomous-driving/adas/

https://www.vector.com/int/en/know-how/technologies/autonomous-driving/adas/

39 © 2020. Vector Informatik GmbH. All rights reserved. Any distribution or copying is subject to prior written approval by Vector. V1.0 | 2020-07-10

Author:Stephan HerzogVector Germany

For more information about Vectorand our products please visit

www.vector.com

http://www.vector.com/

Date post:	07-Oct-2020
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

Online and Offline Validation of ADAS ECUs · 2020. 7. 15. · Online and offline verification by...

Documents