LATIDIA · Robótica

Kinematic MeanFlow: Política de generación de acciones en un solo paso para modelos de cimientos robóticos

arXiv: 2610.00864v1Tipo de anuncio: nuevo Resumen: En este documento, estudiamos cómo lograr la generación de acciones en un solo paso en los Modelos de Fundamentos Robóticos (RFM), con el objetivo de superar la alta latencia de inferencia del flujo de múltiples pasos matc

La noticia

arXiv:2610.00864v1 Announce Type: new Abstract: In this paper, we study how to achieve one-step action generation in Robotic Foundation Models (RFMs), aiming to overcome the high inference latency of multi-step flow matching. MeanFlow provides a promising framework for this goal, yet its direct application leads to performance collapse. We discover that this stems from two distinctive dynamics exhibited in the RFM velocity field: (1) the ``local acceleration" exhibits stability early on, but surges sharply towards the end of the denoising process, and (2) the spread of its magnitudes across samples widens as denoising progresses. To address these issues, we introduce Kinematic MeanFlow (K-MF), a novel one-step action policy tailored for