LATIDIA · Robótica

EmbodiedSmith: Escalar los datos incorporados a través del volante de auto-mejora recursiva en la simulación

arXiv:2610.07969v1 Tipo de anuncio: Cross Resumen: Escalar modelos de cimientos robóticos requiere diversos datos de capacitación y entornos de evaluación confiables. La simulación ofrece una solución escalable, pero la tubería de generación existente

La noticia

arXiv:2610.07969v1 Announce Type: cross Abstract: Scaling robotic foundation models requires diverse training data and reliable evaluation environments. Simulation offers a scalable solution, yet existing generation pipelines remain constrained by predefined assets and skills, a disconnect between scene generation and task generation, and limited support for complex embodiments and physics. We introduce EmbodiedSmith, a framework for scalable embodied data generation through recursive self-improvement (RSI). EmbodiedSmith unifies asset, scene, and task generation in a pipeline that supports autonomous creation and language-driven customization. Its core is an agentic refinement loop: scene generation anticipates downstream task requirements, while task generation guides targeted scene edits, allowing scenes and tasks to iteratively improve one another. This