WorldRFT: 通过强化微调的潜在世界模型进行自动驾驶的规划
WorldRFT: Latent World Model Planning with Reinforcement Fine-Tuning for Autonomous Driving
AI总结：
WorldRFT通过强化学习微调提升自动驾驶规划性能，实现安全性和效率的双重优化。
摘要：
潜在世界模型通过时间自监督学习增强场景表示，提供了一种无需感知标注的端到端自动驾驶方法。然而，以重建为导向的表示学习将感知与规划任务纠缠在一起，导致规划优化不理想。为了解决这一挑战，我们提出了WorldRFT，一种以规划为导向的潜在世界模型框架，通过分层规划分解和局部感知交互细化机制，将场景表示学习与规划对齐，并通过强化学习微调（RFT）来提升安全关键策略性能。具体而言，WorldRFT集成了视觉-几何基础模型以提高3D空间意识，采用分层规划任务分解来引导表示优化，并利用局部感知迭代细化来推导以规划为导向的驾驶策略。此外，我们引入了组相对策略优化（GRPO），它应用轨迹高斯化和碰撞感知奖励来微调驾驶策略，从而在安全方面取得系统性改进。WorldRFT在开放环nuScenes和闭环NavSim基准上实现了最先进的性能。在nuScenes上，它将碰撞率降低了83%（0.30%->0.05%）。在NavSim上，使用仅相机传感器输入时，其性能与基于激光雷达的SOTA方法DiffusionDrive（87.8 vs. 88.1 PDMS）相当。
英文摘要：
Latent World Models enhance scene representation through temporal self-supervised learning, presenting a perception annotation-free paradigm for end-to-end autonomous driving. However, the reconstruction-oriented representation learning tangles perception with planning tasks, leading to suboptimal optimization for planning. To address this challenge, we propose WorldRFT, a planning-oriented latent world model framework that aligns scene representation learning with planning via a hierarchical planning decomposition and local-aware interactive refinement mechanism, augmented by reinforcement learning fine-tuning (RFT) to enhance safety-critical policy performance. Specifically, WorldRFT integrates a vision-geometry foundation model to improve 3D spatial awareness, employs hierarchical planning task decomposition to guide representation optimization, and utilizes local-aware iterative refinement to derive a planning-oriented driving policy. Furthermore, we introduce Group Relative Policy Optimization (GRPO), which applies trajectory Gaussianization and collision-aware rewards to fine-tune the driving policy, yielding systematic improvements in safety. WorldRFT achieves state-of-the-art (SOTA) performance on both open-loop nuScenes and closed-loop NavSim benchmarks. On nuScenes, it reduces collision rates by 83% (0.30% -> 0.05%). On NavSim, using camera-only sensors input, it attains competitive performance with the LiDAR-based SOTA method DiffusionDrive (87.8 vs. 88.1 PDMS).
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
