论文/arXiv 2610.21922·13 天前

TS-WAM：用于视觉语言机器人控制的时序潜在动作与世界-动作耦合

TS-WAM面向视觉语言机器人控制，借助自监督时序学习与未来视觉预测，在仿真与真实操作中验证泛化和鲁棒性。

AI 整理

• 本体 SDK 与控制
• 数据与采集
• 具身模型

Franka FR3

原文链接：TS-WAM: Temporal Latent Actions and World-Action Coupling for Vision-Language Robot Control（arXiv 2610.21922 · 发布日期：2026-10-15）

TS-WAM: Temporal Latent Actions and World-Action Coupling for Vision-Language Robot Control

TS-WAM面向视觉语言机器人控制，借助自监督时序学习与未来视觉预测，在仿真与真实操作中验证泛化和鲁棒性。

核心要点

• 针对依赖动作标注演示且缺少未来视觉状态预测的问题，提升泛化与鲁棒性。
• 提出轻量Teacher-Student World-Action Model，从机器人演示和无动作人类视频中学习。
• 将vlm-jepa与delta-lwm耦合，预测潜在动作和未来视觉特征以指导推理时动作生成。
• 在LIBERO、LIBERO-Plus、SimplerEnv及真实操作实验中表现出强操作性能。

验证范围与局限：验证范围包括LIBERO、LIBERO-Plus、SimplerEnv，以及Franka Research 3和RevoTron真实操作实验。

英文摘要 (Abstract)

Many vision-language-action policies rely on action-labeled robot demonstrations and lack an explicit mechanism for predicting future visual states. These limitations hinder learning from action-free videos and can encourage overfitting to demonstration conditions, weakening generalization and robustness. We present TS-WAM, a lightweight Teacher-Student World-Action Model that addresses these challenges through self-supervised temporal learning and explicit future prediction. Its vlm-jepa framework learns from heterogeneous data, including robot demonstrations and action-free human videos, using an EMA target branch without requiring a separately trained external teacher. By coupling vlm-jepa with delta-lwm, TS-WAM learns to predict latent actions from the current observation and instruction and uses them to predict future visual features that guide action generation at inference. A unified end-to-end training architecture jointly optimizes temporal representation learning, visual dynamics, and robot control without a separate latent-action or world-model pretraining stage. The EMA target branch is removed at inference, keeping the deployed policy lightweight. Experiments on LIBERO, LIBERO-Plus, and SimplerEnv, together with real-world manipulation experiments on Franka Research 3 and RevoTron, demonstrate that TS-WAM achieves strong manipulation performance, generalization, and robustness.

作者：Menghao Yin, Jiaju Yin, Xiyue Dong, Lixin Xu, 张振辉, Ke Zhang, Yating Feng, Renjing Xu

来源

本页事实对应的公开页面，以原文为准。

1. arXiv 2610.21922：TS-WAM: Temporal Latent Actions and World-Action Coupling for Vision-Language Robot Control访问 2026-10-21