LATIDIA · Robótica

ARC: una receta de razonamiento para los modelos de base de robots

arXiv:2610.12386v1 Tipo de anuncio: nuevo Resumen: El enfoque predominante para mejorar los modelos de cimentación de robots (RFM) se basa en modelos más grandes, más demostraciones de robots y capacitación costosa a escala. Mostramos que hay exi

La noticia

arXiv:2610.12386v1 Announce Type: new Abstract: The prevailing approach to improving robot foundation models (RFMs) relies on larger models, more robot demonstrations, and costly training at scale. We show that there exists an effective and efficient complementary approach: the right reasoning recipe can substantially improve the zero-shot task performance of existing state-of-the-art RFMs. We refer to this recipe as ARC. It consists of three key ingredients: a reasoning trace, a scalable automatic labeling pipeline, and a strategy for adapting pretrained RFMs to use these traces for control. First, we find that effective reasoning traces should be grounded in the robot's next action and explain its causal structure: why the action is