Metraipolis › archive › dc_164093cedd23

Resolving State-Representation Mismatch: State-Space Visual Reasoning for Open-Loop VLA Planning — arXiv:2609.33412

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.33412
License (as archived)
CC0-1.0
Archived
2026-10-07T21:01:24.902Z
SHA-256 of the archived text
55a2ac56b625f2c974412cd90440c8f67caa56f72ba17ba3cfc14217fe62e504
Stable URL of this page
https://metraipolis.com/documento/dc_164093cedd23

How to cite

Resolving State-Representation Mismatch: State-Space Visual Reasoning for Open-Loop VLA Planning — arXiv:2609.33412 — arXiv (en), https://arxiv.org/abs/2609.33412 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_164093cedd23 (Metraipolis archive dc_164093cedd23, sha256 55a2ac56b625f2c974412cd90440c8f67caa56f72ba17ba3cfc14217fe62e504)
BibTeX
@misc{dc164093cedd23,
  title = {Resolving State-Representation Mismatch: State-Space Visual Reasoning for Open-Loop VLA Planning — arXiv:2609.33412},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.33412},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_164093cedd23; sha256 55a2ac56b625f2c974412cd90440c8f67caa56f72ba17ba3cfc14217fe62e504},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.33412. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Despite rapid progress in vision-language-action (VLA) models, existing reasoning paradigms still face a fundamental \emph{state-representation mismatch} in open-loop planning. Given only an initial observation, models must internally simulate action-conditioned state transitions, whereas text-, pixel-, and latent-space reasoning can suffer from lossy spatial compression, error-accumulating visual generation, and bypass of intermediate latent tokens, respectively, undermining reliable long-horizon planning. We propose \textbf{State-Space Visual Reasoning} (SSVR), which decouples static visual context, language constraints, and a recurrent latent state. SSVR encodes the initial image and instruction once, then conditions each action prediction on the latent state and updates it with an action-conditioned GRU. Using Qwen2.5-VL as the backbone, SSVR achieves 99.5/99.6, 96.3/98.0, and 83.9/90.6 EM/PR on FrozenLake, Maze, and MiniBehavior, substantially outperforming prior methods. Extensive experiments support the effectiveness of recurrent state modeling for VLA open-loop planning across input transformations and transfer settings. By reusing static visual-textual context and updating a compact recurrent state, SSVR supports efficient multi-step inference, achieving up to $98.58\times$ faster Maze decoding rollouts than the evaluated baselines with the prefix cache prebuilt.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.33412 — 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 2026-10-07T20:29:02.647Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).