LATIDIA · Robótica

Grandes modelos multimodales para el control directo de robots: GPT-6 como punto de partida

arXiv: 2610.28259v1Anuncio Tipo: nuevo Resumen: Los recientes avances en la comprensión visuoespacial y el razonamiento agéntico permiten la manipulación robótica directa con modelos multimodales de propósito general, pero sus capacidades incorporadas

La noticia

arXiv:2610.28259v1 Announce Type: new Abstract: Recent advances in visuospatial understanding and agentic reasoning enable direct robotic manipulation with general-purpose multimodal models, but their embodied capabilities remain poorly understood. This paper systematically evaluate GPT-6 Astra for closed-loop robotic manipulation using a lightweight agent execution framework. We first probe three fundamental embodied capabilities-depth perception, embodied in-context learning, and tactile understanding-and demonstrate that the model can exploit depth information, acquire simple manipulation strategies from contextual demonstrations, and use tactile feedback to correct its actions. We then evaluate GPT-6 Astra across five established manipulation benchmarks along six capability dimensions. Despite competitive performance on several benchmarks-26.92\% success on RoboDojo and 76.0\% on RoboTwin 2.0-our analysis