Dynin-Robotics: Omnimodal Unified Diffusion Vision-Language-Action Model

Hoeun Lee, Jaeik Kim, Jusang Oh +4

Visual goal and dynamics prediction can provide language-conditioned robot policies with both a target outcome and a representation of action-dependent scene changes. We bring thes…