Search papers, labs, and topics across Lattice.
1
15
3
2
Ditch the high-fidelity simulator: IRL-VLA uses a lightweight reward world model trained with inverse reinforcement learning to enable efficient and effective closed-loop RL training for autonomous driving.
