arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2604.14732cs.ROcs.LG
收藏

世界价值-动作模型:面向视觉-语言-动作系统的隐式规划

World-Value-Action Model: Implicit Planning for Vision-Language-Action Systems

Westlake University(西湖大学);Nanjing University Suzhou Campus(南京大学苏州校区)

Runze Li, Hongyin Zhang, Junxi Jin, Qixin Zeng, Zifeng Zhuang, Yiqi Tang, Shangke Lyu, Donglin Wang

AI总结:

本文提出World-Value-Action模型,通过隐式规划提升视觉-语言-动作系统在长时域决策中的性能,通过学习潜在表示和价值函数实现高效规划。

摘要:

视觉-语言-动作(VLA)模型已成为构建具身智能体的有前景范式,但现有方法依赖直接动作预测,缺乏对长时域轨迹的推理和后果评估能力,限制了复杂决策任务的表现。本文引入World-Value-Action(WAV)模型,一种统一框架,使VLA系统能够进行隐式规划。不同于显式轨迹优化,WAV模型学习了基于视觉观察和语言指令的未来轨迹结构化潜在表示。学习的世界模型预测未来状态,而轨迹价值函数评估其长时域效用。动作生成则被公式化为该潜在空间中的推理过程,其中模型逐步将概率质量集中在高价值且动态可行的轨迹上。我们提供了一个理论视角,表明在动作空间中直接规划会导致随着时域增加可行轨迹的概率呈指数衰减。相比之下,潜在空间推理将搜索分布重塑为可行区域,从而实现高效的长时域决策。广泛的仿真和现实世界实验表明,WAV模型在任务成功率、泛化能力和鲁棒性方面均显著优于现有方法,特别是在长时域和组合场景中。代码可在https://github.com/Win-commit/WAV获取。

英文摘要:

Vision-Language-Action (VLA) models have emerged as a promising paradigm for building embodied agents that ground perception and language into action. However, most existing approaches rely on direct action prediction, lacking the ability to reason over long-horizon trajectories and evaluate their consequences, which limits performance in complex decision-making tasks. In this work, we introduce World-Value-Action (WAV) model, a unified framework that enables implicit planning in VLA systems. Rather than performing explicit trajectory optimization, WAV model learn a structured latent representation of future trajectories conditioned on visual observations and language instructions. A learned world model predicts future states, while a trajectory value function evaluates their long-horizon utility. Action generation is then formulated as inference in this latent space, where the model progressively concentrates probability mass on high-value and dynamically feasible trajectories. We provide a theoretical perspective showing that planning directly in action space suffers from an exponential decay in the probability of feasible trajectories as the horizon increases. In contrast, latent-space inference reshapes the search distribution toward feasible regions, enabling efficient long-horizon decision making. Extensive simulations and real-world experiments demonstrate that the WAV model consistently outperforms state-of-the-art methods, achieving significant improvements in task success rate, generalization ability, and robustness, especially in long-horizon and compositional scenarios. Code is available at https://github.com/Win-commit/WAV.

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。

↑