AD technology is shaping fast, with end-to-end AI stacks entering the market quickly. But how to prove it is safe? The UN provided a multipillar approach in its regulation. AD technology progresses as validation struggles to become operational in time. Based on recent commercial and EU work (Hi-Drive, Sunrise, Synergies, etc), TNO will show how to create a consistent safety case, covering both the semantic and the time domain. A practical starting point will be infrastructure coverage. The connection between scenarios, edge cases and evaluating driving behavior will be shown, helping to close the validation gap.

• How to handle the scenario part of the safety case
• How to deal with edge cases in a safety case
• How the infrastructure starting point can help