Reproduce “Robot Learning with Visual Predicted Force” 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
Haonan Chen, Feiyang Wu, Yuxiang Ma, Mustafa Mete, Pengfei Ye, Junxuan Shen, Cheng Zhu, Aurora Ruggeri and 6 more
The method estimates gripping force from gripper deformation in wrist-camera images, annotates demonstrations, and trains a policy to propose action–force pairs. At each step, the robot selects from 32 candidates the action whose proposed force is closest to a demonstration-derived task target. Across four tasks, this improved success over single-sample inference; reorientation remained hardest at 46.7%.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.