Investigating approaches for qualifying virtual testing platforms for automated vehicle certification

2022-01-0982

03/29/2022

Authors Abstract
Content
The safety validation and verification of Automated Driving Systems (ADS) represent a difficult challenge for regulators. Evaluating the performance of an ADS under diverse and complex conditions becomes prohibitive for conventional physical testing. Automotive safety experts are therefore proposing to complement the traditional physical testing with virtual testing to evaluate the performance of ADS in a comprehensive and cost-effective way for a wide range of traffic scenarios. Virtual testing is particularly suited to test ADS under safety critical scenarios that would be difficult and/or unsafe to reproduce on test tracks or public roads. While virtual testing has obvious advantages over the traditional track and real world testing approach, there is a need to ensure that the virtual ADS test results are accurate, representative and can provide confidence in the safety of the ADS, an area which has been less explored. The verification of virtual testing is also required to assess the quality and completeness of the simulations. Starting by selecting the most relevant safety metrics for evaluating performance of simulation-based testing platforms, a method to reliably and accurately compare and quantify the representativeness of the outcomes of virtual testing with real-world testing was developed. The proposed approach underpinned by an Operational Design Domain (ODD) perspective and test objective, was then validated using physical track tests performed at Transport Canada’s Motor Vehicle Test Centre. Based on these findings, a qualification procedure for ADS virtual test platforms was developed as part of the validation process for simulation-based testing. Finally, a plan for a future pilot study involving key stakeholders to demonstrate the feasibility and reproducibility of the approach, and to identify gaps and technical issues is presented. The outcomes of this work are expected to provide a greater understanding of virtual test platforms qualification as a potential validation tool for emerging vehicle technologies and its possible integration to existing vehicle regulations and compliance requirements to enhance road safety.
Meta TagsDetails
Citation
Anctil, B., Khastgir, S., Zhang, J., Burns, P. et al., "Investigating approaches for qualifying virtual testing platforms for automated vehicle certification," SAE Technical Paper 2022-01-0982, 2022, .
Additional Details
Publisher
Published
Mar 29, 2022
Product Code
2022-01-0982
Content Type
Technical Paper
Language
English