Reproduce “ConceptACT: Episode-Level Concepts for Sample-Efficient Robotic Imitation Learning” 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
Jakob Karalus, Friedhelm Schwenker
ConceptACT adds episode-level human concept labels to robot demonstrations, guiding attention to object properties during training. Its policy predicts actions from ordinary visual and robot-state inputs, without requiring labels at deployment. Across sorting and ordering tasks, ConceptACT-Transformer has a reported 0.66 probability of improvement over ACT across training trajectories; this comparison is limited to those tasks.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.