面向端到端自动驾驶的潜变量思维链世界建模
Latent Chain-of-Thought World Modeling for End-to-End Autonomous Driving
把思维链从自然语言搬到与动作同词表的潜空间，并用学习到的潜世界模型 token 表达未来结果，为视觉端到端驾驶提供了一种推理更快、可直接接闭环 RL 后训练的推理时推理范式；其『动作对齐潜空间 + 世界模型 token 交错』的结构可迁移到其他视觉-动作序列决策任务。
原文依据 the model reasons by interleaving (1) action-proposal tokens, which use the same vocabulary as the model's output actions
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论近期面向自动驾驶的视觉-语言-动作（VLA）模型探索在推理阶段进行推理，以提升挑战性场景下的驾驶性能与安全性。多数已有工作使用自然语言来表达产生驾驶动作之前的思维链（CoT）推理。然而，文本未必是推理最高效的表示形式。在本工作中，我们提出 Latent-CoT-Drive（LDrive）：该模型以一种潜变量语言表达思维链，这种语言刻画了正在被考虑的驾驶动作可能产生的结果。我们的方法将思维链推理与决策统一起来，把二者都表示在一个与动作对齐的潜空间中。模型不再使用自然语言，而是通过交错两类 token 进行推理：（1）动作提议 token，其使用与模型输出动作相同的词表；（2）世界模型 token，其以学习到的潜世界模型为基础，表达这些动作的未来结果。我们基于场景的真实未来推演，对模型的动作提议与世界模型 token 进行监督，从而对潜变量思维链进行冷启动。随后我们采用闭环强化学习进行后训练，以增强推理能力。在大规模端到端驾驶基准上，与无推理基线和文本推理基线相比，LDrive 实现了更快的推理速度、更好的轨迹质量，并从交互式强化学习中获得更大的提升。
英文原文摘要 1400 字符
Recent Vision-Language-Action (VLA) models for autonomous driving explore inference-time reasoning as a way to improve driving performance and safety in challenging scenarios. Most prior work uses natural language to express chain-of-thought (CoT) reasoning before producing driving actions. However, text may not be the most efficient representation for reasoning. In this work, we present Latent-CoT-Drive (LDrive): a model that expresses CoT in a latent language that captures possible outcomes of the driving actions being considered. Our approach unifies CoT reasoning and decision making by representing both in an action-aligned latent space. Instead of natural language, the model reasons by interleaving (1) action-proposal tokens, which use the same vocabulary as the model's output actions; and (2) world model tokens, which are grounded in a learned latent world model and express future outcomes of these actions. We cold start latent CoT by supervising the model's action proposals and world model tokens based on ground-truth future rollouts of the scene. We then post-train with closed-loop reinforcement learning to strengthen reasoning capabilities. On a large-scale end-to-end driving benchmark, LDrive achieves faster inference, better trajectory quality, and larger improvements from interactive reinforcement learning compared to both non-reasoning and text-reasoning baselines.