Metraipolis › archive › dc_03521508e6bc

Staircase Policy: Streaming Inference for World-Action Models with Large Action Chunks — arXiv:2609.36471

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.36471
License (as archived)
CC0-1.0
Archived
2026-10-07T16:25:52.345Z
SHA-256 of the archived text
a691f1562a169cab69cf0ca150c879a57ed4aeb66419a735c4bb715eb5b0bc62
Stable URL of this page
https://metraipolis.com/documento/dc_03521508e6bc

How to cite

Staircase Policy: Streaming Inference for World-Action Models with Large Action Chunks — arXiv:2609.36471 — arXiv (en), https://arxiv.org/abs/2609.36471 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_03521508e6bc (Metraipolis archive dc_03521508e6bc, sha256 a691f1562a169cab69cf0ca150c879a57ed4aeb66419a735c4bb715eb5b0bc62)
BibTeX
@misc{dc03521508e6bc,
  title = {Staircase Policy: Streaming Inference for World-Action Models with Large Action Chunks — arXiv:2609.36471},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.36471},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_03521508e6bc; sha256 a691f1562a169cab69cf0ca150c879a57ed4aeb66419a735c4bb715eb5b0bc62},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.36471. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

World-Action Models (WAMs) improve robotic manipulation by conditioning action generation on predicted future observations, but future prediction adds further inference overhead to already expensive iterative action generation. Action chunking can amortize this cost over multiple actions, yet performance degrades over long execution horizons because later actions remain conditioned on stale observations. We introduce STAIRCASE POLICY, a streaming inference and training framework that turns a flow-matching VLA into a JEPA-style WAM and partitions a large action chunk into sub-chunks at staggered denoising stages. Near-term actions are executed as soon as they become available, while later actions continue to be refined. At each sub-chunk boundary, the future latent is re-predicted from the latest observation and used to update all unexecuted actions, enabling long-horizon execution without repeated full policy inference. The resulting future-prediction error can further serve as a signal for adaptive chunking. S-WAM achieves 97.7% on LIBERO and 87.9% on LIBERO-Plus, and improves performance across multiple policy backbones and real-robot tasks. It reaches 292.7 executed actions per second, $3.62\times$ the throughput of conventional execution at comparable accuracy, while reducing time-to-first-action from 123.6 to 73.3 ms. With additional inference optimizations, throughput further increases to 642.9 actions per second.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-30 in cs.AI).
Categorie arXiv: Robotics (cs.RO) ; Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.36471 — 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 2026-10-07T15:54:53.841Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-30?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).