LATIDIA · Robótica

Automejora recursiva en la IA física: mecanismos, evidencia y evaluación.

arXiv:2610.38266v1 Anuncio Tipo: nuevo Resumen: Los sistemas encarnados pueden actualizar políticas, modelos del mundo, habilidades y programas ejecutables a partir de la experiencia de implementación, lo que permite que una interacción beneficie a tareas posteriores. Cuando un sistema

La noticia

arXiv:2610.38266v1 Announce Type: new Abstract: Embodied systems can update policies, world models, skills, and executable programs from deployment experience, allowing one interaction to benefit later tasks. When a system also revises and reuses its data-acquisition, feedback, or learning procedures, experience changes the process that produces capability, introducing recursive self-improvement. This paper surveys research in physical AI through the connection between persistent capability and its improvement process. We first explain the functional distinction between learner updates and improvers, then compare policy learning, imagined experience, memory and skill accumulation, and executable harnesses. These approaches connect to revisions of curricula, rewards, and research programs in different ways. Contact validity, recovery, retention, and resource