Search papers, labs, and topics across Lattice.
CorrectionPlanner is introduced as an autoregressive planning framework for autonomous driving that incorporates an explicit self-correction mechanism based on a propose, evaluate, and correct loop. A learned collision critic predicts potential collisions, and if detected, the planner generates alternative motion tokens conditioned on a "self-correction trace" of past unsafe actions. Trained with imitation learning and model-based reinforcement learning, CorrectionPlanner demonstrates a 20% reduction in collision rate on Waymax and achieves state-of-the-art planning scores on nuPlan.
Autonomous driving planners can now explicitly self-correct unsafe actions by generating motion-token traces conditioned on a learned collision critic, leading to significant safety improvements.
Autonomous driving requires safe planning, but most learning-based planners lack explicit self-correction ability: once an unsafe action is proposed, there is no mechanism to correct it. Thus, we propose CorrectionPlanner, an autoregressive planner with self-correction that models planning as motion-token generation within a propose, evaluate, and correct loop. At each planning step, the policy proposes an action, namely a motion token, and a learned collision critic predicts whether it will induce a collision within a short horizon. If the critic predicts a collision, we retain the sequence of historical unsafe motion tokens as a self-correction trace, generate the next motion token conditioned on it, and repeat this process until a safe motion token is proposed or the safety criterion is met. This self-correction trace, consisting of all unsafe motion tokens, represents the planner's correction process in motion-token space, analogous to a reasoning trace in language models. We train the planner with imitation learning followed by model-based reinforcement learning using rollouts from a pretrained world model that realistically models agents'reactive behaviors. Closed-loop evaluations show that CorrectionPlanner reduces collision rate by over 20% on Waymax and achieves state-of-the-art planning scores on nuPlan.
