Generating useful robot-training data requires more than visually plausiblescenes: objects must support interaction, placements must remain physicallyvalid, and tasks must admit repeatable execution. We present\textbf{Awomo-SimDataEngine}, an agentic system that connects asset and scenegeneration to robot demonstration...
Awomo-PhysicalRSI Team, Dan-Jiao Ma, En-Hui Ma et al.· 0 citations
StrucPhysVideo, a family of video world models that bridges physics-focused data curation with language- and action-conditioned prediction of scene evolution, is presented, and a sparse Mixture-of-Experts text-image-to-video model trained with a curriculum that progressively emphasizes physical dynamics while retaining...
Awomo-WM Team Enhui Ma, Kai-Wen Guo, Ting-Rui Zhang et al.· 0 citations
This work introduces PatchWAM (Patch World-Action Model), which treats continuous actions as another type of patch through a fixed mapping called Action-as-Patch, which allows a single model to predict both how the robot should move and what the scene may look like afterward.
Tian-Heng Wang, Zhou-Yao Xie, Heng-Ji Jia et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
