The increasing complexity of software defined autonomous vehicles, requires new approaches to safety and security validation beyond traditional kilometer-based methods. conventional approaches are time-consuming, costly and create unnecessary CO₂ emissions.

Our presentation demonstrates an innovative validation methodology, based on the strategic collaboration between Yaak Technologies, dSpace and Deloitte, to identify and assess vehicle related safety, security and other risks from internal and external data sources. These risks are assigned in a next step to specific driving scenarios which may be carried out in a virtual environment. Thus, limiting physical test drives and provide assurance for risk mitigations and residual risks.

A key aspect is the relocation of test scenarios to virtual environments, which allows the execution of numerous variations – something not feasible in traditional endurance-testing. With Yaak’s advanced scenario understanding, natural language search capabilities for multimodal data and spatial intelligence models we are able to generate novel driving scenarios from real situations.