RL for Autonomous Driving

- Mid-to-Mid Planning
- Imitation Learning
- Reinforcement Learning

Data-driven AD, Planning

2025

- Mid-to-Mid Autonomous Driving based on Reinforcement Learning (RL)
- Address safety-critical scenarios and long-tail distribution challenges beyond imitation learning (IL) capabilities
- Integrate and convert real-world autonomous driving datasets (nuPlan, Waymo, etc.) into a unified simulation-ready format
- Design reward functions tailored for effective RL training in diverse and rare scenarios

Mid-to-Mid Autonomous Driving based on Reinforcement Learning (RL)

Address safety-critical scenarios and long-tail distribution challenges beyond imitation learning (IL) capabilities
Integrate and convert real-world autonomous driving datasets (nuPlan, Waymo, etc.) into a unified simulation-ready format
Design reward functions tailored for effective RL training in diverse and rare scenarios