
TeraSim is a next-generation digital twin platform designed to evaluate the safety, robustness, and city-scale impact of autonomous vehicles and robots. By integrating real-world maps, adversarial scenario generation, and GenAI-driven visual simulation, TeraSim enables closed-loop testing of autonomy in complex urban environments. The system allows cities and developers to identify unknown unsafe events (UUEs), audit deployment risks, and inform infrastructure and policy decisions through large-scale, behaviorally realistic simulations. It transforms urban areas into intelligent testbeds for governing the autonomous future.
Henry Liu, Howie Sun