MOMASPLAT: Dynamic Novel-View Demonstration Augmentation for Mobile Manipulation
MOMASPLAT: Dynamic Novel-View Demonstration Augmentation for Mobile Manipulation
This research presents MOMASPLAT, a framework designed to augment demonstration data for mobile manipulation by synthesizing trajectories from unseen robot base poses. The method combines 3D Gaussian Splatting and feed-forward 3D foundation models to reconstruct static scenes and manipulated objects in 3D, enabling the generation of geometrically consistent synthetic demonstrations. Evaluated on five real-world tasks involving rigid and articulated objects across three scenes, MOMASPLAT demonstrates improved robustness to navigation uncertainty and localization drift, more than doubling the average success rate of baseline methods while reducing the need for large-scale data collection.
The paper introduces MOMASPLAT, a framework that uses 3D Gaussian Splatting and foundation models to synthesize novel-view demonstrations for mobile manipulation tasks, addressing viewpoint sensitivity in imitation learning by reconstructing both static scenes and dynamic objects.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
number of scenes
3
View original evidence
ted object, render them from novel viewpoints, and generate geometrically consistent synthetic demonstrations paired with transformed actions. We evaluate MOMASPLAT on five real-world mobile manipulation tasks involving both rigid and articulated objects across three scenes, and validate the fidelity of its synthesized demonstrations against ground truth in simulation. MOMASPLAT substantially improves robustness to u
Open source S5
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
proaches use novel-view synthesis (NVS) to augment demonstrations with unseen viewpoints, but they assume static scenes and therefore fail to re-render the manipulated object during the post-contact interaction phase, leaving the most critical parts of the demonstration unaugmented. We present MOMASPLAT, a novel-view demonstration augmentation framework for mobile manipulation that synthesizes full manipulation traje
Open source S3
nseen base poses, more than doubling the average success rate of the strongest baseline and maintaining non-trivial success at large base displacements where all baselines fail, while reducing the need for largescale real-world data collection.
Open source S6
ted object, render them from novel viewpoints, and generate geometrically consistent synthetic demonstrations paired with transformed actions. We evaluate MOMASPLAT on five real-world mobile manipulation tasks involving both rigid and articulated objects across three scenes, and validate the fidelity of its synthesized demonstrations against ground truth in simulation. MOMASPLAT substantially improves robustness to u
Open source S5
Source：arXiv Robotics — research abstracts · arxiv.org