Reasoning to Regulate: Chain-of-Thought for Traffic Rule Understanding
2026-07-27 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors study how self-driving cars can better understand traffic rules by not just recognizing signs but by reasoning about how those signs relate to road lanes and surroundings. They create detailed labels linking each traffic sign to the lanes it controls and improve existing models by teaching them to think step-by-step using a special chain-of-thought method. Their approach includes generating explanations with a language model, checking those explanations for accuracy, and then training the system in two stages to produce better answers. Tests show this reasoning method makes the system more accurate and easier to understand when interpreting traffic regulations.
Autonomous DrivingTraffic Sign RecognitionRegulation ReasoningChain-of-Thought (CoT)Vision-Language Models (VLMs)MapDR DatasetSupervised Fine-Tuning (SFT)Reinforcement Learning (GRPO)Scene ContextLane-level Annotation
Authors
Yueru Luo, Xu Yan, Changqing Zhou, Yiming Yang, Chao Zhan, Shuqi Mei, Chao Zheng, Zhen Li
Abstract
Understanding and complying with traffic regulations is a safety-critical requirement for autonomous driving, yet remains challenging due to the diversity and context dependence of traffic signage. Importantly, regulation understanding is not a simple recognition task, but a reasoning problem: whether a rule applies depends on interpreting the sign in relation to the spatial layout of lanes and scene context. To support such reasoning, MapDR provide fine-grained annotations that link each traffic sign's regulatory rules to the specific lanes they govern. Existing methods, however, largely treat this as direct sequence prediction, ignoring the underlying reasoning that connects sign semantics and map structure. To address this limitation, we explicitly incorporate reasoning into this task and propose a framework that equips vision-language models (VLMs) with chain-of-thought (CoT) capabilities. We first design a scalable CoT curation pipeline that bootstraps rationales from a strong LLM through a two-round strategy and employs a VLM-based verifier to filter out incorrect cases, yielding a high-quality set of (CoT, answer) pairs. Building on this foundation, we adopt a two-stage training scheme: supervised fine-tuning (SFT) to teach rationale-to-answer generation, followed by GRPO reinforcement learning with answer-grounded, fine-grained rewards to further improve final answer accuracy. Extensive experiments on MapDR show that our approach significantly improves both interpretability and accuracy, establishing the first reasoning-based framework for regulation-aware autonomous driving.