Reproduce “PhysiCalWorld: In-Context Physics Calibration for 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
Yunhao Luo, Wentao He, Yinpei Dai, Hyelin Nam, Hongze Fu, Joyce Chai, Nima Fazeli, Jeong Joon Park
PhysiCalWorld helps world models predict scene-specific dynamics by encoding a brief exploratory interaction as a reusable descriptor, without physics labels or test-time fine-tuning. On Push-T, across 15 unseen center-of-mass configurations, DINO-WM with the descriptor achieves 65.1% open-loop IoU and 62.7% planning success, versus 61.4% and 58.7% with raw context. These results are from simulation and assume unchanged scene physics.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.