AI summaryⓘ
The authors studied how teaching autonomous driving models to think step-by-step can be biased when they are shown the actual future paths during training, causing them to make less trustworthy decisions. They found that removing the true future paths stops this bias but makes the task harder. To fix this, they created a new approach where the model picks from a set of possible paths instead of generating one from scratch, making the decisions easier to check. They also developed a method to use the future paths only after the decision is made to improve learning. Their experiments show that these ideas help the model reason better about driving while keeping or improving its ability to understand visual information.
Vision-Language-Action modelsautonomous drivingchain-of-thought supervisiontrajectory anchoring biasfuture trajectorytrajectory generationmultiple-choice questionDEFT-RLVRcausal reasoningmodel verification
Authors
Zixuan Huang, Yang Zhou, Kaixuan Wang, Guli Zhang, Hongyan Xie, Yakun Zhu, Hao Geng, Yikun Ban, Deqing Wang
Abstract
Recent Vision-Language-Action (VLA) models for autonomous driving (AD) increasingly utilize chain-of-thought (CoT) supervision to enhance the reasoning capabilities of their Vision-Language Model (VLM) components, yet existing annotation pipelines commonly expose the teacher model to the logged ground-truth (GT) future trajectory. We empirically show that this induces trajectory anchoring bias: teacher models rationalize the revealed outcome rather than infer a decision from scene evidence, producing less causally faithful CoTs and substantially more severe hallucinations, especially in causally challenging scenes. Removing the GT trajectory eliminates this shortcut, but open-ended trajectory generation entangles high-level decision-making with precise geometric synthesis and low-level dynamics. To make trajectory-level driving decisions verifiable without requiring open-ended trajectory synthesis, we introduce Autonomous-Driving Multiple-Choice Question (AD-MCQ), which casts planning as selection among explicit trajectory candidates. Taking this a step further, we propose Deferred Exposure of Future Trajectories for RLVR (DEFT-RLVR) to transform future trajectories from pre-decision anchors into post-decision verification targets. Experimental results show that DEFT-RLVR improves AD reasoning while preserving or even enhancing general visual capabilities. With VLM-only inference and controllable difficulty through candidate construction, AD-MCQ provides a flexible, scalable, and extensible foundation for future research on verifiable AD reasoning.