arXiv：cs.LG· Songbo Hu, Qiayuan Liao, Yufeng Chi, Kevin Zakka, Yakun Sophia Shao, Pieter Abbeel, Koushil Sreenath·· 4 小时前AI 评分57
Workhorse：从人类数据学习人形机器人全身移动操作
Workhorse: Learning Robust Whole-Body Humanoid Loco-Manipulation from Human Data
AI 导读
arXiv 论文 Workhorse 提出从免机器人的人类演示数据学习人形全身移动操作：视觉规划器预测躯干、双腕、双脚五个连杆目标，强化学习全身跟踪器在机器人上执行，两个策略在同一人类姿态数据上分别训练、无需重定向，并互相模拟对方部署时的误差来增强数据。
正文
Abstract:Humanoid robots still struggle to plan contact-rich whole-body manipulation from egocentric RGB and proprioception. Workhorse learns such manipulation from robot-free human demonstrations. A visual planner predicts five-link targets: the poses of the torso, both wrists, and both feet. A reinforcement-learning whole-body tracker follows them on the robot. Both policies train separately on the same recorded human poses, without retargeting. We augment the training data of each policy to imitate the errors that the other makes at deployment. On a real Unitree G1, Workhorse sorts boxes with its hands and a kick, catches a thrown box, and topples and climbs a suitcase. During box sorting, we show recoveries after a person pushes the robot or takes the box away. In a simulated copy of the demonstration room, the system completes box sorting in 77% of episodes, and in 64% under 40 N.s pushes. With both policies retrained from the same demonstrations, a simulated second humanoid completes box sorting in 83% of episodes without pushes.
| Comments: | 9 pages, 8 figures, 2 tables. Project page: this https URL |
| Subjects: | Robotics (cs.RO); Machine Learning (cs.LG) |
| ACM classes: | I.2.9; I.2.6 |
| Cite as: | arXiv:2610.09117 [cs.RO] |
| (or arXiv:2610.09117v1 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2610.09117 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Songbo Hu [view email]
[v1]
Tue, 6 Oct 2026 21:06:56 UTC (7,412 KB)
来源：arXiv：cs.LG · arxiv.org