WorkDrive: Roadwork Chain of Causation for Autonomous Driving
For autonomous driving systems, this work improves trajectory prediction in roadwork zones, a known bottleneck for VLMs, by introducing causal reasoning.
WorkDrive addresses the failure of autonomous driving VLMs in roadwork zones by constructing perception-grounded causal reasoning aligned with trajectory prediction, achieving a 9.0% reduction in ADE via Chain-of-Causation and a further 3.0% via consistency-based GRPO on the ROADWork dataset.
Autonomous driving vision-language models (VLMs) struggle in roadwork zones, where familiar visual cues such as lane markings and permanent signs are altered or absent, and temporary devices such as cones and barriers redefine the drivable corridor. VLMs can detect these objects, but without explicit guidance they anchor their reasoning on familiar elements from pre-training and fail to connect work-zone observations to correct planning decisions. We propose WorkDrive, a framework that constructs perception-grounded causal reasoning for work zones and aligns it with trajectory prediction. An automated multitask perception pipeline extracts structured scene facts and injects them into a Chain-of-Causation (CoC) annotation pipeline, redirecting the annotator's attention to domain-specific elements. The resulting reasoning labels are used for supervised fine-tuning, followed by reinforcement learning with a single reward: consistency between lateral meta-actions and the predicted trajectory. On ROADWork, the largest public work-zone dataset, the proposed roadwork CoC reduces trajectory average displacement error (ADE) by 9.0\%, and consistency-based GRPO yields a further 3.0\%, achieving progressive improvement over the trajectory-only baseline. Code and data will be publicly released.