ADAS Testing with Virtual Test Drives
DYNA4 is a modular, flexible simulation environment for function development of driver assistance and autonomous driving systems. Testing of ADAS/AD with DYNA4 fulfills the following major requirements on virtual test drives:
Reliable Simulation Models
- Static environment: Road, road signs etc.
- Dynamic environment: traffic participants, animals etc.
- Sensor models: Sense the environment virtually, with different levels of abstraction
- Vehicle dynamics and drivetrain models ensure realistic movement of the vehicle under test including the mounted sensors
Standards for Easy Integrability
- ASAM OpenDRIVE for the road networks
- ASAM OSI sensor interfaces
- ASAM MDF for result data
- FMI for export of DYNA4 models or import of functions
- Unity-based 3D environment, open for extensions and customization
Flexible Execution
- From MIL, SIL up to HIL: From model-based controller development (MIL) to software-in-the-loop (SIL) systems, up to hardware-in-the-loop (HIL) test stands
- On different execution environments, e.g.:
- Simulink's native execution modes
- as a Windows or Linux executable
- as a CANoe model library integrated into a CANoe configuration
- as an FMU integrated into a 3rd-party toolchain
- on many common real-time platforms
- On different scales from single desktop computers to full integration in a CI pipeline
Video: Stimulate ADAS functions in virtual test drives. Benefits of closed-loop simulation with DYNA4 for ADAS function development.
Testing ADAS functions in a virtual world with traffic and environment sensors
DYNA4 GUI with Model Data View and running 3D visualization
Video: How to use virtual test driving in Closed-Loop-Simulations for ADAS testing.
Advantages
- Test comprehensively: Reproducible test driving of thousands of kilometers with parameter variations and test automation
- Fail fast: Lower development costs through correction of malfunctions early in the ADAS/AD development process
- Improve safety: Simulation of critical traffic and collision scenarios without harm
- Reduce efforts: Smooth integration into the Vector toolchain and 3rd-party toolchains
Application Areas
- Environment perception: Lane detection, traffic sign recognition, object detection, simultaneous localization and mapping (SLAM) etc.
- Comfort functions: Adaptive cruise control (ACC), traffic jam assistant, parking pilot etc.
- Safety systems: Pre-crash, front collision warning (FCW), automated emergency braking (AEB), lane keeping assistant (LKA, ELKA), blind spot monitoring etc.
- Car2x: Connected and cooperative driving
Presentation with Demo
Online Presentation with Q&A by Rohde & Schwarz and Vector. HIL Testing Day on 2022-01-20:
Realistic Over-The-Air Validation of Automotive Radar Sensors and ADAS/AD Features
Start a Conversation
Are you interested in virtual test drives for ADAS/AD function development? Then let's talk!
Felix Beygang
Helping to simplify closed-loop simulations. Fom MIL, SIL to HIL.
Models and Components for ADAS Testing
Seamless Integration of Your ADAS Function from MIL to HIL
- Simulink: Integrate your function directly into DYNA4’s Simulink-based models or export DYNA4 models as S-functions for integration into your Simulink model
- C-code: Integrate your function code as S-Functions into the DYNA4 Simulink model or use CANoe’s SIL Adapter Builder and run CANoe simulations with DYNA4 models
- AUTOSAR code: Use vVIRTUALtarget for virtual ECUs and run CANoe simulations with DYNA4 models
- FMU: Integrate your function as an FMU into the DYNA4 model or export the DYNA4 model as an FMU and run it in your preferred environment
- Real ECU: Execute DYNA4 models on a Vector VT System or any other common real-time system in order to test a real ECU on a HIL test stand
显示更多
Sensor Models
- Generic, parameterizable sensor models for radar, lidar, camera and ultrasound
- Standardized sensor data output based on ASAM OSI
- Custom sensor models using ASAM OSI as input
- On object level: object lists, detected lanes, etc.
- On perception level: radar range-doppler map, camera image, segmented camera image, lidar point cloud, segmented lidar point cloud, ultrasound histogram, etc.
- Realistic sensor movement due to physically correct dynamic behavior of the vehicle under test
Read more: Sensor simulation with DYNA4
显示更多
Traffic: Dynamic Environment
- Simulation of surrounding vehicles, pedestrians and animals on complex road networks
- Deterministic tasks to challenge the system under test with specific scenarios based on regulations or expert knowledge
- Stochastic traffic for wider coverage and endurance tests
- Large animation catalog for traffic participants
显示更多
Static Environment
- Native support of the ASAM OpenDRIVE road network model
- Auto-generated visualization of lanes, road signs, traffic signals etc. based on ASAM OpenDRIVE
- Extensible Unity-based visualization with included animation object catalog or customized 3D objects
- Adjustable weather and light conditions like rain, fog, snow, dusk, dawn
显示更多
Vehicle under Test
- Realistic driving dynamics with choice of level of detail
- Passenger cars and commercial vehicles with up to two trailers
- Actuation of engine torque, brakes, steering etc.
- Driving tasks for driver
- Closed-loop interaction with the dynamic and static environment
显示更多
Integration and Workflow Features
ADAS testing with DYNA4: Flexible integration of your SUT and into your work environment
DYNA4 models
- Use DYNA4 models to test your function in closed-loop simulations
- Choose the right level of detail depending on your function inputs and the available parameters
- Integrate your own model components if required
Flexible integration of the System under Test (SUT)
- Use DYNA4’s open interfaces to integrate your function easily from MIL over SIL to HIL
- Execute DYNA4 models in your preferred environment from Simulink over desktop executables to all common HIL platforms
- Choose DYNA4 as the execution master to use the contained workflows or integrate DYNA4 into your existing toolchain
Physics-based sensor models and visualization
- Physics-based sensor models like camera images, lidar point clouds, ultrasonic histograms or range-doppler plot
- Unity-based 3D animation with large object catalog and extensibility with own geometries
- Run-time visualization of simulation results including auto-generated road and terrain with scalable resolution (up to 4k)
Test automation
- Use the integrated task mechanism for comprehensive testing with DYNA4 as the execution master
- Use vTESTstudio for test automation when using CANoe as the execution platform
- Connect external test automation tools via API
Reporting
- Result export in standard MDF format
- Automated report generation
- Extensive results analysis possibilities with Vector tools like vSignalyzer and vMDM
Standards for Virtual ADAS Testing
DYNA4 with native support of typical standards for ADAS testing
In light of increasingly complex distributed vehicle functions, especially for ADAS and autonomous driving, simulation has become a standard tool from function development to validation. But how can virtual test drives be used flexibly and sustainably across disciplines? Standards such as OpenDRIVE, OpenSCENARIO, OSI, FMI, and MDF are important cornerstones in the area of system integration, data management, and test evaluation.
In our technical article "Standards for Virtual Test Drives" (Hanser Automotive Issue 6/2020) we explain how the consistent use of standards enables the reuse of virtual test drives continuously along the entire development process.
