RoboDream: Compositional World Models for Scalable Robot Data Synthesis
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917556552466432 |
|---|---|
| author | Ye, Junjie Xue, Rong Van Hoorick, Basile Li, Runhao Rajaprakash, Harshitha Tokmakov, Pavel Irshad, Muhammad Zubair Guizilini, Vitor Wang, Yue |
| author_facet | Ye, Junjie Xue, Rong Van Hoorick, Basile Li, Runhao Rajaprakash, Harshitha Tokmakov, Pavel Irshad, Muhammad Zubair Guizilini, Vitor Wang, Yue |
| contents | Scaling robot learning requires large-scale, diverse demonstrations, yet real-world data collection via teleoperation remains prohibitively expensive and time-consuming. While video diffusion models offer a promising avenue for data scaling, existing generative approaches are often limited to superficial visual augmentation, or suffer from embodiment hallucinations that yield physically infeasible motions. We present a generalizable embodiment-centric world model that achieves scalable data generation by synthesizing photorealistic demonstrations with novel objects, in novel scenes, and from novel viewpoints. Our approach anchors generation to rendered robot motion while conditioning on explicit scene and object priors, effectively decoupling trajectory execution from environment synthesis. This formulation has the potential to unlock two powerful data scaling capabilities: (1) retrieval and rebirth, which repurposes existing trajectories into entirely new contexts without new motion data; and (2) prop-free teleoperation, where operators manipulate empty air and the model hallucinates the target objects and scene afterwards, eliminating reset time. We demonstrate with real-world experiments that our generated data consistently improves downstream policy performance and significantly reduces real-world data requirements across diverse manipulation tasks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2606_02577 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | RoboDream: Compositional World Models for Scalable Robot Data Synthesis Ye, Junjie Xue, Rong Van Hoorick, Basile Li, Runhao Rajaprakash, Harshitha Tokmakov, Pavel Irshad, Muhammad Zubair Guizilini, Vitor Wang, Yue Robotics Computer Vision and Pattern Recognition Scaling robot learning requires large-scale, diverse demonstrations, yet real-world data collection via teleoperation remains prohibitively expensive and time-consuming. While video diffusion models offer a promising avenue for data scaling, existing generative approaches are often limited to superficial visual augmentation, or suffer from embodiment hallucinations that yield physically infeasible motions. We present a generalizable embodiment-centric world model that achieves scalable data generation by synthesizing photorealistic demonstrations with novel objects, in novel scenes, and from novel viewpoints. Our approach anchors generation to rendered robot motion while conditioning on explicit scene and object priors, effectively decoupling trajectory execution from environment synthesis. This formulation has the potential to unlock two powerful data scaling capabilities: (1) retrieval and rebirth, which repurposes existing trajectories into entirely new contexts without new motion data; and (2) prop-free teleoperation, where operators manipulate empty air and the model hallucinates the target objects and scene afterwards, eliminating reset time. We demonstrate with real-world experiments that our generated data consistently improves downstream policy performance and significantly reduces real-world data requirements across diverse manipulation tasks. |
| title | RoboDream: Compositional World Models for Scalable Robot Data Synthesis |
| topic | Robotics Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2606.02577 |