In safety assurance of automated driving systems, simulation is increasingly being used for critical decisions. The suitability of the used simulation models influence the actual system risk. Therefore, the simulation quality should be assessed systematically, reflected in the safety case, and communicated clearly to involved stakeholders. The following aspects are taken into account:
- Use cases of simulation: Training data, scenario-based testing.
- Communication and value-based development: Communicating the effects of simulation on the system risk with model developers, safety engineers, managers, and authorities.
- Context-sensitive simulation credibility assessment: Derive quality requirements and associated metrics and thresholds from the application context in a traceable manner.
- Formal approaches to the assessment of simulation quality: Design contracts describe the validity domains of simulation models and support re-useability and composition of simulation setups.
- Terminology: Foundation for consistent communication among and across research communities.