HiF-VLA：通过运动表征为视觉-语言-动作模型引入回溯、洞察与前瞻
HiF-VLA: Hindsight, Insight and Foresight through Motion Representation for Vision-Language-Action Models
把帧间运动（而非原始像素）作为紧凑的时序上下文表征，并以此构建以运动为中心的世界模型，为 VLA 提供回溯先验与前瞻推理；这一「运动即上下文」的思路可直接迁移到任何需要长程时序一致性的视觉-动作策略中，且额外推理延迟可忽略。
原文依据 we view motion as a more compact and informative representation of temporal context and world dynamics
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）模型近来通过将视觉与语言线索视觉定位为动作，使机器人操作成为可能。然而，多数 VLA 假设马尔可夫性，仅依赖当前观测，因而受困于时序短视，损害了长程连贯性。在本工作中，我们将运动视为时序上下文与世界动态的一种更紧凑、更具信息量的表征，它捕捉状态间的变化，同时过滤掉静态的像素级噪声。从这一视角出发，HiF-VLA 为 VLA 配备了一个以运动为中心的世界模型，使智能体在动作生成过程中能够对未来的演化进行时序动态推理。基于这一思想，我们提出 HiF-VLA（面向 VLA 的回溯、洞察与前瞻），一个利用运动进行双向时序推理的统一框架。HiF-VLA 通过回溯先验编码过去的动态，通过前瞻推理预判未来的运动，并通过一个回溯调制的联合专家将二者整合，从而为长程操作实现「边想边做」的范式。由此，HiF-VLA 在 LIBERO-Long 与 CALVIN ABC-D 基准上超越了强基线，而额外推理延迟可忽略不计。此外，HiF-VLA 在真实世界长程操作任务中取得显著提升，证明了其在实际机器人场景中的广泛有效性。
英文原文摘要 1395 字符
Vision-Language-Action (VLA) models have recently enabled robotic manipulation by grounding visual and linguistic cues into actions. However, most VLAs assume the Markov property, relying only on the current observation and thus suffering from temporal myopia that degrades long-horizon coherence. In this work, we view motion as a more compact and informative representation of temporal context and world dynamics, capturing inter-state changes while filtering static pixel-level noise. From this perspective, HiF-VLA equips a motion-centric world model for the VLA, enabling agents to reason about temporal dynamics for future evolution during action generation. Building on this idea, we propose HiF-VLA (Hindsight, Insight, and Foresight for VLAs), a unified framework that leverages motion for bidirectional temporal reasoning. HiF-VLA encodes past dynamics through hindsight priors, anticipates future motion via foresight reasoning, and integrates both through a hindsight-modulated joint expert to enable a "think-while-acting" paradigm for long-horizon manipulation. As a result, HiF-VLA surpasses strong baselines on LIBERO-Long and CALVIN ABC-D benchmarks, while incurring negligible additional inference latency. Furthermore, HiF-VLA achieves substantial improvements in real-world long-horizon manipulation tasks, demonstrating its broad effectiveness in practical robotic settings.