Metraipolis › archive › dc_91a44c77ecdb

Correcting WHERE, Preserving HOW: Compositional Generalization for Vision-Language-Action Models via Referential Guidance — arXiv:2609.38616

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.38616
License (as archived)
CC0-1.0
Archived
2026-10-07T13:47:18.946Z
SHA-256 of the archived text
0907e84493e6ac2d56bc6320a07d80e262eade437bdb2ae450ead179a38b0f68
Stable URL of this page
https://metraipolis.com/documento/dc_91a44c77ecdb

How to cite

Correcting WHERE, Preserving HOW: Compositional Generalization for Vision-Language-Action Models via Referential Guidance — arXiv:2609.38616 — arXiv (en), https://arxiv.org/abs/2609.38616 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_91a44c77ecdb (Metraipolis archive dc_91a44c77ecdb, sha256 0907e84493e6ac2d56bc6320a07d80e262eade437bdb2ae450ead179a38b0f68)
BibTeX
@misc{dc91a44c77ecdb,
  title = {Correcting WHERE, Preserving HOW: Compositional Generalization for Vision-Language-Action Models via Referential Guidance — arXiv:2609.38616},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.38616},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_91a44c77ecdb; sha256 0907e84493e6ac2d56bc6320a07d80e262eade437bdb2ae450ead179a38b0f68},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.38616. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

While Vision-Language-Action (VLA) models enable flexible action generation, their generalization across diverse environmental elements, including manipulated objects, destinations, and backgrounds, is limited by the lack of diversity in robotic training data. Trained end-to-end on such data, VLAs tend to exploit visual shortcuts, associating actions with task-irrelevant visual features rather than the intended task semantics. These shortcuts block recomposition of elements already seen by the policy, that is, compositional generalization. Existing approaches mitigate such entanglement through task-relevant perception or targeted data diversification, but offer no explicit mechanism for unseen recomposition and require backbone-specific modifications with retraining. We observe that under such recomposition, VLAs often fail at global grounding while retaining local manipulation skills that recover near the correct target in familiar configurations. Therefore, we propose Referential Guidance (ReGuide), a training-free wrapper that, given object poses from a grounding module, combines semantic and geometric rebinding to guide the end-effector into demonstration-supported configurations of the instructed referent, where the frozen policy can resume execution. Experiments in simulation across multiple VLA backbones as well as on a real robot show that ReGuide improves success rates under compositional shifts by up to 56.8 and 75.0 percentage points, respectively, while preserving standard-task performance.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-01 in cs.LG).
Categorie arXiv: Robotics (cs.RO) ; Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.38616 — 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 2026-10-07T13:18:10.585Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).