论文arXiv 2611.252512 天前
WEVO框架:视频世界模型优化人形VLA真机成功率达75%
本文由 AI 根据公开内容整理,可能有误,请以原文为准:WEVO: World Model-Enabled VLA Optimization for Whole-Body Loco-Manipulation(arXiv 2611.25251)
核心内容
针对人形机器人强化学习中真机探索不安全及仿真Sim-to-Real间隙问题,研究团队发布WEVO框架。该方法无需遥操作数据,通过在动作条件视频世界模型(WM)中进行VLA后训练,结合仿真Play Data与少量真机自主Rollout对齐模型。实验表明,在三个真实机器人任务中,引入视频WM的RL策略将平均成功率从基线23.3%大幅提升至75.0%,远超纯真机RL的48.3%。
技术架构上,WEVO由VLA、全身控制器(WBC)与视频WM组成闭环RL环境并提供密集进度奖励。仿真生成的操作与运动Play Data用于增强接触动力学与视角变化预测能力;仅需每任务20次真机自主Rollout即可实现WM与部署观测的对齐。该方案互补利用了仿真的策略初始化广度与视频WM的部署一致性,项目代码与演示已在GitHub开源。
关键看点
- 性能跃升:视频世界模型RL将三任务平均成功率从23.3%提至75.0%,优于纯真机RL的48.3%
- 数据高效:每任务仅需20次真机自主Rollout即可完成世界模型与真实部署环境的对齐
- 闭环架构:集成VLA、全身控制器WBC与视频WM构建RL环境,提供密集进度奖励信号
- 免遥操作:利用仿真Play Data增强接触动力学预测,摆脱对昂贵人工遥操作数据的依赖
工程与产业观察
WEVO验证了视频世界模型作为“部署对齐模拟器”在人形机器人VLA后训练中的关键价值。相比传统Sim-to-Real,该方案以极低真机交互成本(20次Rollout)换取显著性能增益,为具身智能在缺乏海量真机数据场景下的策略迭代提供了可复现的工程范式,尤其适合全身控制等高维复杂任务的落地验证。
英文摘要 (Abstract)
Reinforcement learning (RL) can improve vision-language-action (VLA) policies beyond imitation, but humanoid exploration is unsafe and simulator RL can exploit residual sim-to-real gaps. We present WEVO, a teleoperation-free framework for post-training a whole-body humanoid VLA in an action-conditioned video world model (WM). Simulated manipulation and locomotion play data improve contact-dynamics and viewpoint-change prediction. A VLA trained on synthetic demonstrations collects 20 autonomous real robot rollouts per task to align the WM with deployment observations. The VLA, whole-body controller (WBC), and video WM then form a closed-loop RL environment with dense progress rewards. Across three real robot tasks, play data improves every evaluated WM metric and downstream policy success; video WM RL raises mean success from 23.3% to 75.0%, compared with 48.3% for real robot RL. We therefore leverage the complementary strengths of both environments: simulation provides policy initialization and broad coverage, while the video WM offers a deployment-aligned environment for policy improvement. Project website: https://qianzhong-chen.github.io/wevo.github.io/.
作者:Qianzhong Chen, Muyuan Lin, Yuyin Sun, Mac Schwager, Cheng-Hao Kuo, Roberto Mart'in-Mart'in
外部来源
本页事实对应的公开页面,以原文为准。(点击将离开本站)