Metraipolis › archive › dc_500ed74ce220
Back on Track: Deployment-time Policy Improvement via World Model Planning — arXiv:2610.19579
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.19579
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-15T21:57:52.715Z
- SHA-256 of the archived text
ad120caf1850870c2ba322fd3bbd90a1e0ff5d9476331f23ba59ea1b74767699- Stable URL of this page
https://metraipolis.com/documento/dc_500ed74ce220
How to cite
Back on Track: Deployment-time Policy Improvement via World Model Planning — arXiv:2610.19579 — arXiv (en), https://arxiv.org/abs/2610.19579 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_500ed74ce220 (Metraipolis archive dc_500ed74ce220, sha256 ad120caf1850870c2ba322fd3bbd90a1e0ff5d9476331f23ba59ea1b74767699)
BibTeX
@misc{dc500ed74ce220,
title = {Back on Track: Deployment-time Policy Improvement via World Model Planning — arXiv:2610.19579},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.19579},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_500ed74ce220; sha256 ad120caf1850870c2ba322fd3bbd90a1e0ff5d9476331f23ba59ea1b74767699},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.19579. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
State-of-the-art visuomotor policies are increasingly capable of executing complex tasks, yet can fail during deployment in ways they cannot predict or recover from on their own. Improving their performance and robustness typically requires collecting more data, retraining, or modifying the policy architecture, all of which are time- and resource-intensive in practice. We instead propose an offline-trained monitoring-and-intervention framework that improves a policy during deployment using only black-box query access. From a limited set of labeled on-policy rollouts, we learn a policy-conditioned liveness value that measures the nominal policy's progress toward task completion; we additionally train a compact latent world model using these rollouts together with play data. At deployment, the liveness value continuously monitors execution. When it detects a loss of progress, a short-horizon planner searches the world model for corrective actions using the same value as its planning objective, after which control returns to the nominal policy. We evaluate our method on the LIBERO-Goal task suite and on a Franka Panda manipulator, using the vision-language-action model $\pi_{0.5}$ and a diffusion policy as nominal policies, and demonstrate significant improvements in task success rates, while supporting real-time control. Additional results and videos are available on the project website ( this https URL ).
Tema degli esploratori: statistica e scienza dei dati (elenco arXiv cs.RO, nuovi lavori del giorno).
Categorie arXiv: Robotics (cs.RO).
Pagina del lavoro: https://arxiv.org/abs/2610.19579 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-15T21:38:44.458Z dalla pagina https://arxiv.org/list/cs.RO/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).