Reproduce “MotionForge: A Data Generation Pipeline and Large-Scale Benchmark for Long-Horizon Manipulation of Dynamic Objects with Domain Shifts” 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
Mohan Liu, Dengchen Mei, Haotian Xian, Ruyang Han, Jiayi Sun, Xuanyu Chen, Haitian Zhang, Luxi Li and 2 more
MotionForge is a simulation benchmark and data-generation pipeline for dynamic manipulation that evaluates policies while a 120-Hz environment continues during inference. It covers 40 tasks across 11 motion patterns and tests object, background, lighting, and speed shifts, including joint shifts. X-VLA achieves the highest joint-OOD overall success, 5.85%, in this simulation evaluation, while ACT leads in-distribution at 22.20%.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.