LATIDIA · Robótica

M$^3 $ P-R1: Aprendizaje de refuerzo para la planificación de movimiento multimodal guiada por un modelo de lenguaje grande a través de la generación de código MIP

arXiv:2609.18669v1 Tipo de anuncio: nuevo Resumen: La planificación de movimiento multimodal (M$^3 $P) requiere un razonamiento conjunto sobre movimientos continuos y transiciones de modo discreto, lo que dificulta su resolución de manera eficiente. Por ejemplo,

La noticia

arXiv:2609.18669v1 Announce Type: new Abstract: Multi-Modal Motion Planning (M$^3$P) requires joint reasoning over continuous motions and discrete mode transitions, making it difficult to solve efficiently. For instance, a bipedal robot may walk to a target location and then use its arms to grasp an object. This scenario captures both mode transitions and continuous dynamics, yielding feasible paths that neither purely discrete nor continuous planners can handle. While Mixed-Integer Programming (MIP) offers a principled framework, constructing tractable formulations for non-convex problems is typically manual and domain-specific, especially in the approximate, discretization-based MIP regime needed for non-convex robotic tasks. We propose M$^3$P-R1, a reinforcement learning method that fine-tunes large language models (LLMs)