Rui Gan

The results show that reliable scenario generation requires verifying realized behavior, not merely executable code, and demonstrate the value of explicit intermediate representations for verifiable and repairable language-driven co-simulation.

Despite advances in vision-language models, they still struggle to reason about the complex dynamics of traffic crashes, highlighting a critical gap in safety-critical applications.

Autonomous vehicles can now better navigate complex, rare events by explicitly incorporating long-tail risk cues into end-to-end trajectory planning using a novel multimodal driving framework.