Scenario Generation for Testing Automated Driving Systems

The SOTIF-standard (ISO 21448) establishes scenario-based testing as state-of-the-art for verifying and validating Advanced Driver Assistance Systems (ADAS) and Automated Driving Systems (ADS). However, the SOTIF standard lacks details for selecting scenarios used as test cases. Consequently, missing details of SOTIF hinder its practical application. In this paper, we analyze existing scenario generation techniques and discuss whether they generate SOTIF-compliant scenario suites. Subsequently, we leverage variability modeling techniques to address two essential challenges that remain open: How to model an overall scenario space and how to cover it practically? We elaborate on sampling strategies and coverage criteria that are relevant for generating SOTIF-compliant scenario suites. Finally, we assess and compare generated scenario suites using mutation testing indicating the ability of scenario suites to detect potential failures.