Reproduce “Hierarchical Residual Policy Learning for Real-World Mobile Manipulation with Sparse Human Guidance” 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
Zhefei Gong, Hefeifei Jiang, Zehao Wang, Jiachen Li
HiRe-MoMa adapts frozen mobile-manipulation policies by learning bounded residual corrections, split between base and arm heads, with the arm conditioned on the base correction. Across three TIAGo++ tasks, equally weighted task scores rise from 39.2% to 75.0%, while episode length falls from 455.0 to 352.7 steps. Pick Place remains hardware-limited, and evidence covers one platform.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.