aiqbench analysis
This arXiv paper proposes LoopWAM, a robot policy that learns from the full range of deployment rollouts, including failures and suboptimal behavior, rather than only expert demonstrations. Its reported gains (23.67% to 43.33% success over three rounds on RoboChallenge Table30-V2) come from benchmark results, not real-world deployment at scale, so readers should check how well the closed-loop approach holds up across tasks and hardware. The most transferable idea is the joint training of action generation and action-conditioned video world modeling.
LoopWAM: Closed-Loop Generalist Policy Improvement from Deployment Rollouts
arXiv:2610.18873v1 Announce Type: new Abstract: Scaling expert demonstrations alone does not enable generalist robot policies to improve from deployment. Real-world rollouts contain successes, suboptimal behavior, and failures: imitating every recorded action can reinforce poor behavior, while filtering out failures discards evidence of physical interactions and their consequences. We introduce \textbf{LoopWAM}, a closed-loop world-action model that learns directly from this full quality spectrum, using rollouts from both itself and other deployed policies. Source-dependent noise sampling unifies behavior learning and action-conditioned world modeling, while quality steering and quality-weighted supervision retain full video supervision across qualities and regulate action imitation. Video and action tokens interact throughout parallel denoising, allowing learned consequences to inform action generation directly. On RoboChallenge Table30-V2, three deployment-and-training rounds raise success from 23.67\% to 43.33\%, a 19.7-percentage-point gain, with a final score of 56.80. LoopWAM also achieves 94.5\% average success on RoboTwin~2.0 and 49.5\% overall success on RoboCasa365 without additional embodied pretraining, obtaining the best overall results among the compared methods across all three benchmarks. On challenging shirt folding task, two rounds of deployment-driven improvement after single-task post-training raise success from 13.20\% to 55.56\%. These results demonstrate closed-loop improvement from mixed-quality experience under both multi-task learning and single-task post-training. \href{https://loop-wam.github.io/}{\textcolor{orange}{Project Page}}
Abstract:Scaling expert demonstrations alone does not enable generalist robot policies to improve from deployment. Real-world rollouts contain successes, suboptimal behavior, and failures: imitating every recorded action can reinforce poor behavior, while filtering out failures discards evidence of physical interactions and their consequences. We introduce LoopWAM, a closed-loop world-action model that learns directly from this full quality spectrum, using rollouts from both itself and other deployed policies. Source-dependent noise sampling unifies behavior learning and action-conditioned world modeling, while quality steering and quality-weighted supervision retain full video supervision across qualities and regulate action imitation. Video and action tokens interact throughout parallel denoising, allowing learned consequences to inform action generation directly. On RoboChallenge Table30-V2, three deployment-and-training rounds raise success from 23.67% to 43.33%, a 19.7-percentage-point gain, with a final score of 56.80. LoopWAM also achieves 94.5% average success on RoboTwin 2.0 and 49.5% overall success on RoboCasa365 without additional embodied pretraining, obtaining the best overall results among the compared methods across all three benchmarks. On challenging shirt folding task, two rounds of deployment-driven improvement after single-task post-training raise success from 13.20% to 55.56%. These results demonstrate closed-loop improvement from mixed-quality experience under both multi-task learning and single-task post-training. Project page: this https URL
Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
arXiv:2610.18873 [cs.RO]
(or arXiv:2610.18873v2 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2610.18873 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 297 words
Related coverage
Learning While Deploying: Fleet-Scale Reinforcement Learning for Generalist Robot Policies
arXiv · Aug 25, 2026
Learning While Deploying: Fleet-Scale Reinforcement Learning for Generalist Robot Policies
arXiv · Jun 4, 2026
The Machines that Make the Machines
NVIDIA · Oct 7, 2026
Runway’s WorldPrompt and the Engineering of Real-Time Worlds
Latent Space · Sep 25, 2026