Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2605.03637 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909014469640192 |
|---|---|
| author | Li, Zhiyuan Yang, Wenyan Zhao, Wenshuai Ma, Yue Tu, Yuanpeng Marttinen, Pekka Pajarinen, Joni |
| author_facet | Li, Zhiyuan Yang, Wenyan Zhao, Wenshuai Ma, Yue Tu, Yuanpeng Marttinen, Pekka Pajarinen, Joni |
| contents | Learning robotic manipulation from human videos is a promising solution to the data bottleneck in robotics, but the distribution shift between humans and robots remains a critical challenge. Existing approaches often produce entangled representations, where task-relevant information is coupled with human-specific kinematics, limiting their adaptability. We propose a generative framework for cross-embodiment video editing that directly addresses this by learning explicitly disentangled task and embodiment representations. Our method factorizes a demonstration video into two orthogonal latent spaces by enforcing a dual contrastive objective: it minimizes mutual information between the spaces to ensure independence while maximizing intra-space consistency to create stable representations. A parameter-efficient adapter injects these latent codes into a frozen video diffusion model, enabling the synthesis of a coherent robot execution video from a single human demonstration, without requiring paired cross-embodiment data. Experiments show our approach generates temporally consistent and morphologically accurate robot demonstrations, offering a scalable solution to leverage internet-scale human video for robot learning. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_03637 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Bridging the Embodiment Gap: Disentangled Cross-Embodiment Video Editing Li, Zhiyuan Yang, Wenyan Zhao, Wenshuai Ma, Yue Tu, Yuanpeng Marttinen, Pekka Pajarinen, Joni Robotics Learning robotic manipulation from human videos is a promising solution to the data bottleneck in robotics, but the distribution shift between humans and robots remains a critical challenge. Existing approaches often produce entangled representations, where task-relevant information is coupled with human-specific kinematics, limiting their adaptability. We propose a generative framework for cross-embodiment video editing that directly addresses this by learning explicitly disentangled task and embodiment representations. Our method factorizes a demonstration video into two orthogonal latent spaces by enforcing a dual contrastive objective: it minimizes mutual information between the spaces to ensure independence while maximizing intra-space consistency to create stable representations. A parameter-efficient adapter injects these latent codes into a frozen video diffusion model, enabling the synthesis of a coherent robot execution video from a single human demonstration, without requiring paired cross-embodiment data. Experiments show our approach generates temporally consistent and morphologically accurate robot demonstrations, offering a scalable solution to leverage internet-scale human video for robot learning. |
| title | Bridging the Embodiment Gap: Disentangled Cross-Embodiment Video Editing |
| topic | Robotics |
| url | https://arxiv.org/abs/2605.03637 |