As the industry adopts physical AI, traditional testing alone cannot solve safety edge cases. This presentation discusses how data automation moves beyond collection. It will demonstrate how unifying real-world logs with hyper-scale simulation identifies critical scenarios and unknown unknowns and how it can increase V&V efficiency by 10x, providing a repeatable path to commercializing safe, scalable autonomous systems.

• Unifying real-world and virtual data: a road and simulation feedback loop
• Automated scenario curation: identifying data from driving hours
• Hyper-scale synthetic generation: handling unknowns using edge cases