Work

Human-Like Interactive Lane-Change Modeling Based on Reward-Guided Diffusive Predictor and Planner