10:32官方账号arXiv cs.AI@Cong Xu, Ravi SankarA study on a decision layer for autonomous driving that uses a language-guided intent module to prevent planning failures due to intent misinterpretation. The module gates planned maneuvers before commitment, with successful trajectory control in all ten replays. False triggers were reduced by treating uncertainty as abstention, and the model's contribution was confirmed by ablations that bound its performance.论文intent misinterpretationautonomous drivingdecision layer推荐理由:Read this paper if you're interested in how AI can prevent planning failures in autonomous driving by anticipating intent divergence and using a gating mechanism.原文稍后读已读值得跟进有用关注 intent misinterpretation