Reproduce “Physical Diffusion Policy via MPM-Guided 3D Gaussian Simulation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Can Wang, Qiyuan Qiao, Lei Liu, Ge Yuan, Xinhui Liu, Dong Xu
The authors estimate object physics from video and use those properties to condition a material-aware diffusion policy for manipulation. The system links 3D Gaussian Splatting with Material Point Method simulation and combines four material experts. In real-robot elastic-object pick-and-place, it reports 57.1% success versus 30.0% for the DP3 baseline; this category-specific result does not show gains across all tasks or objects.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.