Determining critical driving scenarios is crucial in testing autonomous vehicle functions. Safety-critical functions need to be exposed to the limit cases in order to verify and validate their safety in automated driving. The methodology for determining these critical driving scenarios is not a straight forward task because of the highly unpredictable situations that appear in urban traffic. For the development of ADAS and such components, detailed standard-based approaches exist in the literature and practice. However, for virtual safety validation, a de facto standard has not yet been adopted.
Therefore, we are actively working on efficient simulation-based testing of ADAS and ADF. The workflow that we have adopted is an intuitive one and tries to determine the boundary of conditions where failures start occuring. In the following figure below, we have illustrated the chosen approach for our testing and benchmarking. An academic article about this guided simulation approach for determining the safety levels of autonomous vehicle components and functions will soon be published, and the relevant link will be added to the "Publications and Articles" section.