TRM-VLA：面向视觉-语言-动作模型的时序感知思维链推理与记忆
TRM-VLA: Temporal-Aware Chain-of-Thought Reasoning and Memorization for Vision-Language-Action Models
TRM-VLA 将时序感知的思维链推理与记忆机制引入 VLA 模型，其关键帧触发推理（KTR）和粒度自适应上下文记忆（GCM）可被视觉研究者复用于需要长程规划与实时执行的机器人操作任务，为视觉-语言-动作模型的推理效率与一致性提供新思路。
原文依据 TRM-VLA consists of two core components: (1) Keyframe-Triggered Reasoning (KTR), which identifies task progress and performs hierarchical CoT reasoning only at key decision points to reduce redundant inference
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论视觉-语言-动作（VLA）模型已成为通用机器人操作的一种强大范式。然而，现有方法通常省略中间推理步骤并直接回归动作，限制了推理的可解释性以及在长程或组合任务中的性能。尽管近期研究将思维链（CoT）推理引入VLA模型，但由于两个关键问题，其效果仍不理想：（1）在每个时间步生成完整的推理轨迹会引入大量冗余，从而阻碍实时部署；（2）推理独立进行，忽略了时序一致性，导致规划冲突。我们提出TRM-VLA，一个时序感知的推理与记忆框架，将显式时序建模集成到VLA推理过程中。TRM-VLA包含两个核心组件：（1）关键帧触发推理（KTR），它识别任务进度并仅在关键决策点执行分层思维链推理，以减少冗余推理；（2）粒度自适应上下文记忆（GCM），它动态存储和检索历史推理轨迹，以保持帧间连贯性和全局上下文。基于双系统架构——结合用于慢速推理的多模态基础模型（系统2）和用于快速执行的扩散策略（系统1）——TRM-VLA学习以统一方式高效规划和行动。在LIBERO-90、SIMPLER以及四个真实世界机器人任务上的大量实验表明，TRM-VLA在提升推理效率的同时达到了最优性能。
英文原文摘要 1645 字符
Vision-Language-Action (VLA) models have emerged as a powerful paradigm for general robotic manipulation. However, existing approaches typically omit intermediate reasoning steps and directly regress actions, limiting reasoning interpretability and performance in long-horizon or compositional tasks. Although recent studies introduce Chain-of-Thought (CoT) reasoning into VLA models, their effectiveness remains suboptimal due to two key issues: (1) generating a full reasoning trajectory at every timestep introduces substantial redundancy, thereby hinders real-time deployment and (2) reasoning is performed independently, neglecting temporal consistency, which leads to planning conflicts.We propose TRM-VLA, a temporal-aware reasoning and memorization framework that integrates explicit temporal modeling into the VLA reasoning process. TRM-VLA consists of two core components: (1) Keyframe-Triggered Reasoning (KTR), which identifies task progress and performs hierarchical CoT reasoning only at key decision points to reduce redundant inference; and (2) Granularity-adaptable Context Memory (GCM), which dynamically stores and retrieves historical reasoning trajectories to maintain inter-frame coherence and global context. Built upon a dual-system architecture--combining a multimodal foundation model for slow reasoning (System 2) with a diffusion-based policy for fast execution (System 1)--TRM-VLA learns to plan and act efficiently in a unified manner. Extensive experiments on LIBERO-90, SIMPLER, and four real-world robotic tasks demonstrate that TRM-VLA achieves state-of-the-art performance while improving reasoning efficiency.