Metraipolis › archive › dc_cd83ce30a0d2

Perceives but Doesn't Brake: A Perception-Action Binding Gap in Reasoning Driving VLAs — arXiv:2610.28184

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.28184
License (as archived)
CC0-1.0
Archived
2026-10-21T07:08:18.053Z
SHA-256 of the archived text
be0d7684dcaa26e5c05fb67f32dfc6e0bb2baf7e19e4362adc5445694f084fb3
Stable URL of this page
https://metraipolis.com/documento/dc_cd83ce30a0d2

How to cite

Perceives but Doesn't Brake: A Perception-Action Binding Gap in Reasoning Driving VLAs — arXiv:2610.28184 — arXiv (en), https://arxiv.org/abs/2610.28184 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cd83ce30a0d2 (Metraipolis archive dc_cd83ce30a0d2, sha256 be0d7684dcaa26e5c05fb67f32dfc6e0bb2baf7e19e4362adc5445694f084fb3)
BibTeX
@misc{dccd83ce30a0d2,
  title = {Perceives but Doesn't Brake: A Perception-Action Binding Gap in Reasoning Driving VLAs — arXiv:2610.28184},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.28184},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cd83ce30a0d2; sha256 be0d7684dcaa26e5c05fb67f32dfc6e0bb2baf7e19e4362adc5445694f084fb3},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.28184. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Reasoning-augmented vision-language-action (VLA) driving models can perceive and reason about safety-critical hazards yet fail to translate that evidence into safe actions. We characterize this failure as a perception-action binding gap: hazard information is available inside the model but exerts insufficient control over trajectory generation. Using matched simulated counterfactuals and logged driving replay, we show that in-path hazards are actually encoded in the visual representation, but longitudinal plans are dominated by ego-speed history and respond only weakly to hazard proximity or the model's own reasoning trace. Further analysis shows that the trajectory generator is capable of braking but sometimes fails to do so due to the perception-action binding gap. Without retraining or modifying any weights, we inject a hazard-conditioned energy into the frozen flow-matching sampler of Alpamayo-1.5, restoring the missing coupling at generation time. With exact obstacle geometry, guidance cuts footprint intrusion from 58.1% to 14.5% and recovers safe braking whenever sufficient stopping distance remains, eliminating collisions entirely in our braking-relevant closed-loop battery. It also reverses the diagnosed dependence of the plan, from tracking ego speed to responding to obstacle range. These results separate hazard perception from hazard control in driving VLAs, show that unsafe behavior can arise even when the model has both the evidence and the ability to act on it, and propose a simple fix for bridging this binding gap.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Robotics (cs.RO) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.28184 — 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 2026-10-21T07:05:43.628Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).