Salvato in:
| Autori principali: | , , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2509.11839 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866918397536632832 |
|---|---|
| author | Liu, Jiacheng Ding, Pengxiang Zhou, Qihang Wu, Yuxuan Huang, Da Peng, Zimian Xiao, Wei Zhang, Weinan Yang, Lixin Lu, Cewu Wang, Donglin |
| author_facet | Liu, Jiacheng Ding, Pengxiang Zhou, Qihang Wu, Yuxuan Huang, Da Peng, Zimian Xiao, Wei Zhang, Weinan Yang, Lixin Lu, Cewu Wang, Donglin |
| contents | Recent Vision-Language-Action models show potential to generalize across embodiments but struggle to quickly align with a new robot's action space when high-quality demonstrations are scarce, especially for bipedal humanoids. We present TrajBooster, a cross-embodiment framework that leverages abundant wheeled-humanoid data to boost bipedal VLA. Our key idea is to use end-effector trajectories as a morphology-agnostic interface. TrajBooster (i) extracts 6D dual-arm end-effector trajectories from real-world wheeled humanoids, (ii) retargets them in simulation to Unitree G1 with a whole-body controller trained via a heuristic-enhanced harmonized online DAgger to lift low-dimensional trajectory references into feasible high-dimensional whole-body actions, and (iii) forms heterogeneous triplets that couple source vision/language with target humanoid-compatible actions to post-pre-train a VLA, followed by only 10 minutes of teleoperation data collection on the target humanoid domain. Deployed on Unitree G1, our policy achieves beyond-tabletop household tasks, enabling squatting, cross-height manipulation, and coordinated whole-body motion with markedly improved robustness and generalization. Results show that TrajBooster allows existing wheeled-humanoid data to efficiently strengthen bipedal humanoid VLA performance, reducing reliance on costly same-embodiment data while enhancing action space understanding and zero-shot skill transfer capabilities. For more details, For more details, please refer to our \href{https://jiachengliu3.github.io/TrajBooster/}. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_11839 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | TrajBooster: Boosting Humanoid Whole-Body Manipulation via Trajectory-Centric Learning Liu, Jiacheng Ding, Pengxiang Zhou, Qihang Wu, Yuxuan Huang, Da Peng, Zimian Xiao, Wei Zhang, Weinan Yang, Lixin Lu, Cewu Wang, Donglin Robotics Computer Vision and Pattern Recognition Recent Vision-Language-Action models show potential to generalize across embodiments but struggle to quickly align with a new robot's action space when high-quality demonstrations are scarce, especially for bipedal humanoids. We present TrajBooster, a cross-embodiment framework that leverages abundant wheeled-humanoid data to boost bipedal VLA. Our key idea is to use end-effector trajectories as a morphology-agnostic interface. TrajBooster (i) extracts 6D dual-arm end-effector trajectories from real-world wheeled humanoids, (ii) retargets them in simulation to Unitree G1 with a whole-body controller trained via a heuristic-enhanced harmonized online DAgger to lift low-dimensional trajectory references into feasible high-dimensional whole-body actions, and (iii) forms heterogeneous triplets that couple source vision/language with target humanoid-compatible actions to post-pre-train a VLA, followed by only 10 minutes of teleoperation data collection on the target humanoid domain. Deployed on Unitree G1, our policy achieves beyond-tabletop household tasks, enabling squatting, cross-height manipulation, and coordinated whole-body motion with markedly improved robustness and generalization. Results show that TrajBooster allows existing wheeled-humanoid data to efficiently strengthen bipedal humanoid VLA performance, reducing reliance on costly same-embodiment data while enhancing action space understanding and zero-shot skill transfer capabilities. For more details, For more details, please refer to our \href{https://jiachengliu3.github.io/TrajBooster/}. |
| title | TrajBooster: Boosting Humanoid Whole-Body Manipulation via Trajectory-Centric Learning |
| topic | Robotics Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2509.11839 |