LATIDIA · Robótica
Palabras físicas: un vocabulario para estructurar el razonamiento latente en la manipulación de robots
arXiv: 2610.26854v1Tipo de anuncio: nuevo Resumen: El razonamiento multimodal proporciona modelos de cimientos robóticos con el contexto predictivo necesario para la generación de acciones confiables. Sin embargo, el razonamiento basado en el lenguaje captura la semanalidad de la tarea
WhatsApp ↗Telegram ↗
La noticia
arXiv:2610.26854v1 Announce Type: new Abstract: Multimodal reasoning provides robotic foundation models with predictive context needed for reliable action generation. However, language-based reasoning captures task semantics but struggles to express fine-grained physical dynamics, while latent reasoning models future state evolution but offers limited structure for organizing these dynamics into reusable reasoning primitives. To bridge this gap, we introduce PhysWord, a reasoning-before-acting foundation model built on Physical Words, which use atomic action verbs as shared semantic anchors for continuous representations of physical dynamics. Specifically, PhysWord predicts upcoming atomic action words and directly aligns their contextual hidden states with latent targets derived from the corresponding future observations. Building on these primitives, we construct