LATIDIA · Robótica

Hacia la era de la inteligencia encarnada agénica

arXiv: 2611.09124v1Tipo de anuncio: nuevo Resumen: Los modelos de cimentación guían cada vez más las acciones de los robots, pero la autonomía sostenida requiere un control eficiente y la recuperación de fallas. Presentamos A2-MEAS, un fr multi-agente incorporado

La noticia

arXiv:2611.09124v1 Announce Type: new Abstract: Foundation models increasingly guide robot actions, yet sustained autonomy requires efficient control and failure recovery. We introduce A2-MEAS, an embodied multi-agent framework built around code as a scaling interface. Learned correction codes combine corrective language, spatial evidence, execution history, and action proposals. MetaBrain provides training supervision and runtime guidance. A correction encoder and adapter absorb this knowledge into policy weights during training, while fresh codes adapt behavior with frozen weights at deployment. The action model, A2-WLA, adds motion-aware history and a latent world model to a vision-language-action policy. Surprisingly, this simple design with action pretraining and task specialization reaches 0.5755 Q-score on the 50-task BEHAVIOR