Metraipolis › archive › dc_ebfae6e5f90c
Continue or Replan? Bernoulli-Continuation Policy Learning for Adaptive Horizon Execution — arXiv:2608.03483
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.03483
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-31T12:19:23.910Z
- SHA-256 of the archived text
85e1ea2b7b2508d90d7772538a1eee664b9ed023ee87388b1f2a4099e2e0dac8- Stable URL of this page
https://metraipolis.com/documento/dc_ebfae6e5f90c
How to cite
Continue or Replan? Bernoulli-Continuation Policy Learning for Adaptive Horizon Execution — arXiv:2608.03483 — arXiv (en), https://arxiv.org/abs/2608.03483 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ebfae6e5f90c (Metraipolis archive dc_ebfae6e5f90c, sha256 85e1ea2b7b2508d90d7772538a1eee664b9ed023ee87388b1f2a4099e2e0dac8)
BibTeX
@misc{dcebfae6e5f90c,
title = {Continue or Replan? Bernoulli-Continuation Policy Learning for Adaptive Horizon Execution — arXiv:2608.03483},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.03483},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ebfae6e5f90c; sha256 85e1ea2b7b2508d90d7772538a1eee664b9ed023ee87388b1f2a4099e2e0dac8},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.03483. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Existing chunk-based Vision-Language-Action (VLA) models execute a fixed number of actions (i.e., execution horizon) before replanning, turning replanning into a task-agnostic periodic schedule that is independent of task progress. As a result, when no replanning boundary falls before a critical manipulation stage, it is executed from a stale chunk rather than a freshly replanned one. To address this limitation, we propose Bernoulli-Continuation Policy (BCP), a lightweight, plug-and-play framework for adaptive horizon execution that keeps the base VLA frozen. Given a fixed-length action chunk, its continuation head decomposes execution-horizon selection into a sequence of continue-or-replan decisions, which imposes an ordinal, prefix-sharing inductive bias over candidate horizons rather than treating them as independent classes. Since the optimal horizon for each chunk is not observable, we train this head with reinforcement learning from trajectory-level outcomes and introduce a Replanning-Efficiency Reward that jointly rewards task success and efficient VLA usage, discouraging the policy from collapsing to unnecessarily short horizons. On RoboTwin 2.0 with LingBot-VLA as the base policy, BCP improves the average success rate by +11.08% on 13 low-success tasks and from 89.88% to 93.94% (+4.06%) across all 50 tasks. Although trained only under the Clean setting, BCP generalizes to the Randomized setting, raising the average success rate by +4.06%. It also transfers to a different base policy $\pi_{0.5}$, achieving a better result on LIBERO (+1.7%) and, notably, on the harder LIBERO-PRO (+6.8%). On a real robot, BCP lifts success from 74% to 92% and from 44% to 84% on two manipulation tasks. Meanwhile, its negligible overhead, combined with higher success, makes BCP's overall runtime even lower than the fixed-horizon baselines.
Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-05 in cs.AI).
Categorie arXiv: Robotics (cs.RO) ; Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2608.03483 — 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 2026-10-31T12:05:53.050Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-05?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).