LATIDIA · Robótica

RoboFoundry: evolución del sistema como política para agentes incorporados en el autoaprendizaje

arXiv:2609.32862v1 Tipo de anuncio: nuevo Resumen: Un modelo de cimentación no debe actuar de forma aislada como un agente incorporado. Sin embargo, los métodos existentes a menudo optimizan los componentes individuales de la pila de agentes, como la memoria, el contexto,

La noticia

arXiv:2609.32862v1 Announce Type: new Abstract: A foundation model should not act in isolation as an embodied agent. Yet, existing methods often optimize individual components of the agent stack, such as memory, context, skills, or action interfaces, rather than treating the supporting system itself as a unified policy. Moreover, interaction alone does not yield self-improvement unless execution experience is converted into persistent, validated system changes. We therefore propose RoboFoundry, the first embodied agentic framework that formulates this process as Self-Evolving System-as-Policy. RoboFoundry diagnoses capability gaps in decision-making and memory management, converts execution traces into validated task-specific system updates, and promotes recurring improvements to the general system. Evolution operates over two complementary