InfGen: Scenario Generation as Next-Token-Group Prediction

InfGen is a unified transformer that treats the entire driving scene—map, lights, agent states and motions—as one long token stream and rolls it out autoregressively.
Generates both initial states, traffic light signals, and traffic participant trajectories in a single model.
Supports dynamic agent injection for open-world, long-horizon simulation.
Autonomous driving planners trained in InfGen scenarios show ↑ robustness & generalization over log-replay baselines.