Newest papers

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

Matteo Cercola, Katie Z Luo, Simone Formentin, Marco Pavone · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.28184 · cited by 0 works (OpenAlex) · open-access PDF

Korean

이 실패는 인식-행동 결합 격차로 특징화됩니다 : 위험 정보는 모델 내에서 사용할 수 있지만 경로 생성에 대한 충분한 제어를 행사하지 않습니다. 일치 된 시뮬레이션 반사 사실과 기록된 운전 재생을 사용하여, 우리는 경로 위험이 실제로 시각적 표현에 인코딩되지만 장거리 계획은 자아 속도 역학에 의해 지배되며 위험 가까이에만 약하게 반응합니다. 더 많은 분석은 경로 생성자가 브레이크 할 수 있지만 때로는 인식-행동 결합 격차로 인해 그렇게 할 수 없음을 보여줍니다. 심지어는 브레이크 계획을 재훈련하거나 수정하지 않으면 위험 조건의 에너지를 위험 조절 에너지 흐름에 삽입하면 알파미아오이오

Machine translation (Meta M2M100); the English original follows.

Abstract

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.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Trajectory, Computer science, Hazard, Bayesian probability, Control theory (sociology), Obstacle

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange