Newest papers

Reasoning as a Distribution: Generative Plan Latents for Vision-Language-Action Models

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

DOI: 10.48550/arxiv.2610.28166 · cited by 0 works (OpenAlex) · open-access PDF

Korean

Vision-Language-Action (VLA) 정책은 행동하기 전에 점점 더 논리적이지만, 기존 논리 모델은이 중간 결정을 하나의 표현으로 압축합니다. 여러 계획이 유효할 때,이 약속은 다른 방법으로 선택되거나 수정 될 수있는 대안을 숨깁니다. 우리는 행동하기 전에 논리적 상태를 계획에 대한 낮은 차원 조건적 분배로 나타내는 Generative Plan Latent (GPL)를 도입합니다. 이것은 논리적 상태를 해결할 수 있도록합니다 : VLA를 다시 훈련하지 않고, 계획은 샘플링 할 수 있습니다 (커버레이션), 요청에 따라 선택 (제어), 그리고 배포 결과 (결과 적응 테스트)에 의해 다시 무게됩니다. 우리는 $π_0$와 같은 SimplerEnv 조작에 세 가지를 사용하고

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

Abstract

Vision-Language-Action (VLA) policies increasingly reason before acting, but existing reasoning models compress this intermediate decision into a single representation. When several plans are valid, this commitment hides alternatives that could otherwise be selected or revised. We introduce the Generative Plan Latent (GPL), which represents the reasoning state as a low-dimensional conditional distribution over plans, such as which object, grasp, or route to use. This makes the reasoning state addressable: without retraining the VLA, plans can be sampled (coverage), selected on request (steering), and reweighted by deployment outcomes (outcome adaptation). We test all three on SimplerEnv manipulation with $π_0$, and extend the analysis to RoboCasa, LIBERO, and a maze domain. Steering the plan raises the fraction of executions that follow the request from 0.48 to 0.94, while the same guidance on the native action distribution shows no reliable gain. Finally, we show that an explicit plan is most useful when it carries decision information not already determined by the observation.

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

Topics: Computer science, Artificial intelligence, Plan (archaeology), Generative grammar, Retraining, Machine learning

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