Robotic Spatiotemporal Editing: A Unified Interface for Controllable Action Generation via Masked Action Reconstruction
Robotic Spatiotemporal Editing: A Unified Interface for Controllable Action Generation via Masked Action Reconstruction
This research introduces the MAP (Masked Action Space) interface to unify spatiotemporal control references for robot manipulation. The proposed Inpainting Policy (IP) reconstructs missing action dimensions using progress-guided alignment and multi-mask training, achieving an average task success rate of 92.67% on ManiSkill and 64.67% on LIBERO-10. Real-world experiments show average positional and rotational errors of 44.3 mm and 7 degrees, respectively, validating the policy's ability to guide executable robot behaviors using diverse references.
The authors propose a unified interface for controllable robot action generation by treating spatiotemporal references as partial specifications of an action space, achieving high task success rates in simulation and demonstrating low error metrics in real-world experiments.
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
Reported success rate
92.67%
Reported trials
View original evidence
ing unseen action-constraint combinations. We further adopt a training-free RePaint variant that iteratively inpaints the action during denoising to reduce control error. On ManiSkill and LIBERO-10, our policy achieves average task success rates of 92.67% and 64.67%, respectively.Real-world experiments demonstrate spatiotemporal control with average positional and rotational errors of 44.3 mm and 7{\deg}, respectivel
Open source S4Reported success rate
64.67%
Reported trials
View original evidence
ing unseen action-constraint combinations. We further adopt a training-free RePaint variant that iteratively inpaints the action during denoising to reduce control error. On ManiSkill and LIBERO-10, our policy achieves average task success rates of 92.67% and 64.67%, respectively.Real-world experiments demonstrate spatiotemporal control with average positional and rotational errors of 44.3 mm and 7{\deg}, respectivel
Open source S4positional error
44.3
View original evidence
ing unseen action-constraint combinations. We further adopt a training-free RePaint variant that iteratively inpaints the action during denoising to reduce control error. On ManiSkill and LIBERO-10, our policy achieves average task success rates of 92.67% and 64.67%, respectively.Real-world experiments demonstrate spatiotemporal control with average positional and rotational errors of 44.3 mm and 7{\deg}, respectivel
Open source S4rotational error
7
View original evidence
ing unseen action-constraint combinations. We further adopt a training-free RePaint variant that iteratively inpaints the action during denoising to reduce control error. On ManiSkill and LIBERO-10, our policy achieves average task success rates of 92.67% and 64.67%, respectively.Real-world experiments demonstrate spatiotemporal control with average positional and rotational errors of 44.3 mm and 7{\deg}, respectivel
Open source S4
- paper: not_reported
What remains unknown
Not established in the collected evidence: Environment, Control, Data origin.
Reported performance applies to the described task. It does not establish general autonomy or deployment readiness.
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
ing unseen action-constraint combinations. We further adopt a training-free RePaint variant that iteratively inpaints the action during denoising to reduce control error. On ManiSkill and LIBERO-10, our policy achieves average task success rates of 92.67% and 64.67%, respectively.Real-world experiments demonstrate spatiotemporal control with average positional and rotational errors of 44.3 mm and 7{\deg}, respectivel
Open source S4
Source：arXiv Robotics — research abstracts · arxiv.org