Metraipolis › archive › dc_49cb3349d15f
DIVA: Dual-Space Intent-Aware Visual Attenuation for Vision-Language-Action Policies — arXiv:2610.09144
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.09144
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-08T11:09:00.749Z
- SHA-256 of the archived text
a29bf44d8bd6930b2e288e2e74ada4101c3a1c5fedfd10e30a7858c6e7a06f8e- Stable URL of this page
https://metraipolis.com/documento/dc_49cb3349d15f
How to cite
DIVA: Dual-Space Intent-Aware Visual Attenuation for Vision-Language-Action Policies — arXiv:2610.09144 — arXiv (en), https://arxiv.org/abs/2610.09144 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_49cb3349d15f (Metraipolis archive dc_49cb3349d15f, sha256 a29bf44d8bd6930b2e288e2e74ada4101c3a1c5fedfd10e30a7858c6e7a06f8e)
BibTeX
@misc{dc49cb3349d15f,
title = {DIVA: Dual-Space Intent-Aware Visual Attenuation for Vision-Language-Action Policies — arXiv:2610.09144},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.09144},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_49cb3349d15f; sha256 a29bf44d8bd6930b2e288e2e74ada4101c3a1c5fedfd10e30a7858c6e7a06f8e},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.09144. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Vision-language-action (VLA) policies typically feed dense visual patch tokens into a language-action backbone, preserving scene context but offering no explicit mechanism to regulate how strongly different visual tokens influence policy computation. We introduce DIVA, a Dual-Space Intent-Aware Visual Attenuation module with an anchor-then-attenuate design. DIVA combines high-level task intent with low-level visual evidence to estimate patch-wise relevance anchors, then applies them in two complementary spaces: it reweights projected visual tokens before backbone entry and persistently attenuates low-relevance visual states within the backbone. DIVA preserves the full visual token sequence and requires no external grounding supervision. On LIBERO, DIVA improves OpenVLA-OFT from 96.6% to 98.0% average success and raises its zero-shot LIBERO-Plus score from 69.6 to 72.6. Real-world experiments further show consistent gains under task-irrelevant visual perturbations, supporting the robustness of intent-aware visual attenuation beyond simulation. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2610.09144 — 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 2026-10-08T10:14:05.909Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).