FactorDrive: Adaptive Multi-Step Reasoning Driven by Planning-Critical Factors for End-to-End Autonomous Driving

2026-08-10

AI Summary

Existing end-to-end autonomous driving approaches struggle to effectively integrate spatial-physical evidence into planning reasoning, often resulting in rigid inference pathways and a lack of explicit optimization for planning quality. To address this, this work proposes an adaptive multi-step reasoning framework grounded in Planning-Critical Factors (PCFs). The authors construct a PCF-CoT dataset to enable fine-grained, scene-adaptive generation of reasoning trajectories and introduce, for the first time, the QS-GRPO algorithm, which synergistically combines Monte Carlo Tree Search with trajectory-level planning rewards to jointly optimize reasoning pathways and driving policies. Coupled with large-scale driving instruction fine-tuning, the proposed method achieves state-of-the-art open-loop and closed-loop planning performance on both the nuScenes and NAVSIM benchmarks.

Abstract

Vision-language models (VLMs) have advanced scene understanding and enabled explicit reasoning in end-to-end autonomous driving. However, existing methods insufficiently integrate spatial-physical evidence into planning reasoning, while reasoning adaptation remains coarse-grained and falls short of scene-specific planning demands. Furthermore, reasoning-path optimization for higher planning quality remains largely unexplored in autonomous-driving post-training. To address these limitations, we propose FactorDrive, an end-to-end autonomous driving framework for adaptive multi-step reasoning driven by planning-critical factors (PCFs). We first perform large-scale driving-domain instruction tuning to establish foundational driving knowledge. Building on this foundation, we construct PCF-CoT, a chain-of-thought (CoT) dataset that grounds planning reasoning in trajectory-relevant spatial-physical evidence and organizes reasoning around scene-specific PCFs, enabling the composition and depth of reasoning paths to adapt to different planning demands. We further introduce Quality Search-Guided Group Relative Policy Optimization (QS-GRPO), which guides Monte Carlo Tree Search (MCTS) with trajectory-level planning rewards to discover reasoning paths with higher planning quality and uses the resulting responses to optimize the policy through GRPO, thereby improving trajectory planning performance. Extensive experiments on both open-loop (nuScenes) and closed-loop-oriented (NAVSIM) benchmarks demonstrate that FactorDrive achieves state-of-the-art planning performance.

Problem

• autonomous driving
• planning reasoning
• spatial-physical evidence
• reasoning adaptation
• trajectory planning

Innovation

• planning-critical factors
• adaptive multi-step reasoning
• chain-of-thought
• trajectory-level reward
• policy optimization