外观
论文arXiv 2609.307351 个月前
Praxis：从第一视角视频中蒸馏物理交互先验用于可泛化全身操控
原文链接：Praxis: Distilling Physical Interaction Priors from Egocentric Videos for Generalizable Whole-Body Manipulation（arXiv 2609.30735 · 发布日期：2026-09-25）
Praxis: Distilling Physical Interaction Priors from Egocentric Videos for Generalizable Whole-Body Manipulation
Praxis从单次第一视角视频提取交互先验，结合姿态校准与在线感知，在五个长时程操控任务中验证泛化。
核心要点
- 移动人形操控需在物体姿态和接触变化下兼顾工作空间与精细手物交互。
- Praxis结合单次第一视角示范的物理交互先验、闭环姿态校准和在线感知。
- 框架协调目标导航、臂手工作空间对齐，以及上下肢同步的灵巧操控三阶段。
- 每项技能仅需一个人类示范且无需任务特定重训，实验显示多种泛化与扰动恢复。
验证范围与局限：验证范围为摘要所述五个长时程操控任务。
英文摘要 (Abstract)
Mobile humanoid manipulation requires both reaching a usable workspace and preserving precise hand-object interactions as object poses and contact conditions change. Learning these behaviors from limited task-specific data remains challenging. To bridge this gap, we introduce Praxis, a whole-body manipulation framework that combines physical interaction priors from one-shot egocentric video demonstrations with closed-loop posture calibration and online perception. The framework coordinates three stages: vision-language-guided navigation toward target objects, closed-loop posture calibration to align the arm-hand workspace, and dexterous manipulation with synchronized upper- and lower-body control. Online visual feedback re-grounds demonstrated interaction geometry under new object poses and scene configurations, while tactile feedback adapts hand motions to actual contact conditions. Each manipulation skill is specified by one human demonstration, without task-specific manipulation-policy retraining. Experiments across five long-horizon manipulation tasks demonstrate spatial, visual, and cross-object generalization, as well as recovery from external physical disturbances across all three stages.
作者：Shuliang He, Ruiyan Xu, Bo Yue, Hengming Zhang, Huayi Zhou, Shuai Wang, Wei-Shi Zheng, Guiliang Liu
来源
本页事实对应的公开页面，以原文为准。