Reproduce “MoRE: Mixture-of-Experts for Multi-Physics Robotic World Models” 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
Ziyu Zhu, Shaowei Liu, Albert J. Zhai, Henry Che, Wei-Chiu Ma, Shenlong Wang
MoRE predicts contact-driven object motion by routing a short interaction history to specialists for rigid pushing, deformable pushing or sliding, and deformable grasping. In DROID’s deformable non-prehensile evaluation, it reports RMSE of 0.014, compared with 0.028 for PointWorld, across manipulated-object points rather than whole-scene flow. The model focuses on one manipulated object and does not represent fine-grained physical properties.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.