GUARD: Anomaly-Informed Q-Steering for Generative Robot Policies in Dexterous Manipulation
GUARD: Anomaly-Informed Q-Steering for Generative Robot Policies in Dexterous Manipulation
This research presents GUARD, a framework that leverages anomaly scores derived from training data to guide generative robot policies during execution.
The paper introduces GUARD, a framework that uses anomaly scores from training data to steer generative robot policies during execution, achieving high success rates on Adroit dexterous tasks and demonstrating sim-to-real transfer without modifying the pretrained policy.
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.
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
preventing trajectories from entering poorly supported regions, without directly modifying the pretrained policy. Across four Adroit dexterous manipulation tasks, GUARD achieves average success rates of $96.0\%$ and $80.0\%$ under nominal and perturbed dynamics, outperforming the strongest baseline by $21.0$ and $20.0$ percentage points while maintaining the lowest anomaly score. GUARD further demonstrates direct si
Open source S4
m-to-real transfer on in-hand screwdriver rotation and reorientation tasks, achieving $40$ and $25$ percentage-point gains in success rate. These results demonstrate that anomaly signals can serve not only for failure monitoring but also to complement sparse task rewards as effective supervision for robust policy steering.
Open source S5
Source：arXiv Robotics — research abstracts · arxiv.org