For the virtual validation of ADAS functions and autonomous vehicles, a large number of scenario simulations is needed. We present an innovative software solution that uses large language models to automatically create simulation models quickly and efficiently, either from textual descriptions and functional requirements or from recorded video and other data, extracting the relevant features and parameters. Corner cases can be detected and the scenario criticality increased to support the safety argumentation for the homologation.

• ADAS/AD virtual validation
• Scenario creation using generative AI
• ADAS/AD testing process acceleration