Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2502.16587 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908655088041984 |
|---|---|
| author | Xie, Sicheng Cao, Haidong Weng, Zejia Xing, Zhen Chen, Haoran Shen, Shiwei Leng, Jiaqi Wu, Zuxuan Jiang, Yu-Gang |
| author_facet | Xie, Sicheng Cao, Haidong Weng, Zejia Xing, Zhen Chen, Haoran Shen, Shiwei Leng, Jiaqi Wu, Zuxuan Jiang, Yu-Gang |
| contents | Distilling knowledge from human demonstrations is a promising way for robots to learn and act. Existing methods, which often rely on coarsely-aligned video pairs, are typically constrained to learning global or task-level features. As a result, they tend to neglect the fine-grained frame-level dynamics required for complex manipulation and generalization to novel tasks. We posit that this limitation stems from a vicious circle of inadequate datasets and the methods they inspire. To break this cycle, we propose a paradigm shift that treats fine-grained human-robot alignment as a conditional video generation problem. To this end, we first introduce H&R, a novel third-person dataset containing 2,600 episodes of precisely synchronized human and robot motions, collected using a VR teleoperation system. We then present Human2Robot, a framework designed to leverage this data. Human2Robot employs a Video Prediction Model to learn a rich and implicit representation of robot dynamics by generating robot videos from human input, which in turn guides a decoupled action decoder. Our real-world experiments demonstrate that this approach not only achieves high performance on seen tasks but also exhibits significant one-shot generalization to novel positions, objects, instances, and even new task categories. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_16587 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Human2Robot: Learning Robot Actions from Paired Human-Robot Videos Xie, Sicheng Cao, Haidong Weng, Zejia Xing, Zhen Chen, Haoran Shen, Shiwei Leng, Jiaqi Wu, Zuxuan Jiang, Yu-Gang Robotics Distilling knowledge from human demonstrations is a promising way for robots to learn and act. Existing methods, which often rely on coarsely-aligned video pairs, are typically constrained to learning global or task-level features. As a result, they tend to neglect the fine-grained frame-level dynamics required for complex manipulation and generalization to novel tasks. We posit that this limitation stems from a vicious circle of inadequate datasets and the methods they inspire. To break this cycle, we propose a paradigm shift that treats fine-grained human-robot alignment as a conditional video generation problem. To this end, we first introduce H&R, a novel third-person dataset containing 2,600 episodes of precisely synchronized human and robot motions, collected using a VR teleoperation system. We then present Human2Robot, a framework designed to leverage this data. Human2Robot employs a Video Prediction Model to learn a rich and implicit representation of robot dynamics by generating robot videos from human input, which in turn guides a decoupled action decoder. Our real-world experiments demonstrate that this approach not only achieves high performance on seen tasks but also exhibits significant one-shot generalization to novel positions, objects, instances, and even new task categories. |
| title | Human2Robot: Learning Robot Actions from Paired Human-Robot Videos |
| topic | Robotics |
| url | https://arxiv.org/abs/2502.16587 |