Reproduce “LiMoDE: Rethinking Lifelong Robot Manipulation from a Mixture-of-Dynamic-Experts Perspective” 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
Zhihao Gu, Lin Wang
LiMoDE lets robots adapt to manipulation tasks by adding task-specific low-rank experts while keeping earlier experts available for selection and combination. During pretraining, a router uses visual temporal differences as a proxy for how many heterogeneous experts to activate. On ten-task LIBERO-LONG, it reports higher FWT, lower BWT, and higher AUC than a PPL reimplementation, though gains are not universal.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.