The role of ASAM standards in scenario-based testing: Insights from BMW and Aptiv

Summary

ASAM OpenX standards improve interoperability and efficiency for the Simulation and testing of automated driving systems. OpenX standards, such as ASAM OpenDRIVE, ASAM OpenCRG, ASAM OpenSCENARIO, and ASAM OSI (Open Simulation Interface), are instrumental for scenario-based testing. This article will discuss how leading automotive organizations adopt these standards for simulating virtual worlds, recreating scenarios from real-world data, and validating ADAS ECUs.
 

What is scenario-based testing? How does ASAM support it?

Scenario-based testing involves creating specific situations or scenarios that one might encounter while operating advanced driver assistance systems (ADAS)/autonomous driving (AD) systems. ASAM standards play an important role in scenario-based testing by facilitating interoperability, data exchange, and toolchain integration. ASAM OpenDRIVE and ASAM OpenCRG are used to describe and exchange road networks between simulators. ASAM OpenDRIVE is used to describe elements of a scene such as roads, lanes, junctions, traffic signs, etc. ASAM OpenCRG helps in describing road surfaces that are used for testing vehicle dynamics. Scenarios involve dynamic elements, such as the ego vehicle (the vehicle equipped with ADAS/AD systems) and other traffic participants moving within the scene. ASAM OpenSCENARIO is used as a File Format to describe the dynamic content of the driving and traffic simulators. ASAM OSI provides compatibility between automated driving functions and driving Simulation frameworks.

MathWorks supports ASAM standards in its software for automated driving, including RoadRunner and Automated Driving Toolbox. This support has helped its customers address specific challenges in the development, testing, and Validation of ADAS/AD technologies. Each section below represents an example project where ASAM standards were a critical part of the workflow.
 

Virtual world generation for BMW driving simulation


BMW’s advanced driving Simulation center, featuring 14 distinct simulators, is at the forefront of
testing and validating new vehicle functionalities, with a particular focus on ADAS and UI/UX
concepts. The center creates large, accurate road networks for Simulation studios, using software
tools such as Spider, Unreal Engine, and RoadRunner. ASAM standards, such as ASAM OpenDRIVE
and ASAM OpenCRG, are used to provide consistency, accuracy, and interoperability across
different Simulation setups.

One challenge BMW faced was to ensure that the 3D models of roads and environments accurately
matched the logical descriptions provided by its Simulation software, Spider. By leveraging ASAM
OpenDRIVE, BMW could ensure a seamless match between the 3D visual content and the logical
road network data. BMW used RoadRunner to import ASAM OpenDRIVE files and then generate
corresponding 3D models for a consistent and accurate representation of virtual environments.

Another challenge for BMW was to accurately recreate various road surfaces for use
in simulating and testing vehicle dynamics with high precision. ASAM OpenCRG, an open
tshtuaarnfta ddcaersdc i rnib deest aroila, den abled BMW to simulate real and synthetic road conditions with accuracy. BMW
used RoadRunner to edit ASAM OpenCRG files and adjust road surfaces in its scenarios. By using
ASAM standards, such as ASAM OpenDRIVE and ASAM OpenCRG, BMW ensured consistency
between Simulation logic and 3D visualization, preventing discrepancies in virtual environment.
BMW has presented this work at the MathWorks Automotive Conference in Germany, and more
details can be found here: Virtual World Generation for BMW Driving Simulation

Harvesting driving scenarios from recorded sensor data

Aptiv wanted to perform regression testing of its AD/ADAS algorithms by harvesting driving
scenarios from recorded sensor data. Data from GPS, IMU, camera, radar, and lidar sensors were
fused to accurately reconstruct driving scenarios such as lane changes, vehicle cut-ins, and more.
This involved steps such as ego and non-ego vehicle trajectory reconstruction, scenario generation,
and Validation.

One challenge Aptiv faced was to export these scenarios for use in other simulators and ensure the
behavior remains the same. Aptiv used RoadRunner to export scenarios to the ASAM
OpenSCENARIO XML format and test them in the esmini tool. The ASAM OpenSCENARIO Standard
played a crucial role in ensuring the generated scenarios could be exported and run across various
Simulation tools for AD/ADAS software testing. Aptiv has presented this work at the 2023
MathWorks Automotive Conference in Novi, Michigan. 

Leveraging ASAM OSI for ADAS ECU validation

Aptiv wanted to efficiently validate and test its ADAS/AD systems without the extensive need for
real-world road tests. Its approach was to create a virtual Simulation environment that replicates
real-world conditions, which is then used to simulate and test vehicle dynamics, controls, and sensor
accuracy. Creating the virtual environment involved extracting and processing various objects, such
as moving and stationary objects, traffic signs, and environmental information, and transforming
these from world coordinates to sensor coordinates for detailed analysis.
ASAM OSI plays a crucial role in Aptiv’s Simulation ecosystem by serving as a standardized
middleware that facilitates the exchange of information between environmental simulators and its
sensor and logic models. This Standard helps in improving interoperability between automated
driving functions and driving Simulation frameworks.

RoadRunner was used for creating Aptiv’s virtual Simulation environments. RoadRunner Scenario
along with MATLAB were used to simulate and export ground truth sensor data into the ASAM OSI
File Format. These files were used to validate Aptiv’s sensor and logic models. ASAM OSI ensured
integration and interoperability between different environmental simulators and models without
requiring changes to the interface. Aptiv’s work on the ASAM OSI using RoadRunner for ADAS ECU
Validation was presented at the MathWorks Automotive Conference held in Chennai, India in 2023.

Conclusion

ASAM standards have emerged as key enablers in the area of autonomous driving, offering a unified
framework for Simulation and testing. MathWorks worked with leading automotive companies such
as BMW and Aptiv in the adoption of these standards by significantly improving their Simulation
capabilities. This standardization not only promotes interoperability but also accelerates the
development and Validation of safer, more efficient ADAS/AD systems.