Reproduce “UMI-Live for In-The-Wild Bimanual Robotics Data Collection” 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
Govind Velamoor, Adrian Lin, Sam Jung, Spencer Mateega, Brandon Y. Yang, Amish Sethi
UMI-Live coordinates two sensor-equipped handheld grippers for bimanual demonstrations and processes recordings asynchronously as collection continues. Its pipeline reconstructs motion and prepares synchronized datasets for robot training and deployment. On a six-orange-to-a-bowl task with dual YAM arms, MolmoAct2 succeeds in 8/10 trials after fine-tuning on 75% UMI and 25% robot-teleoperation data, versus 0/10 for its released checkpoint. This result does not isolate UMI's contribution.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.