Reproduce “Identifying Important Sensory Feedback for Learning Locomotion Skills” 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
Wanming Yu, Chuanyu Yang, Christopher McGreavy, Eleftherios Triantafyllidis, Guillaume Bellegarda, Milad Shafiee, Auke Jan Ijspeert, Zhibin Li
The authors rank feedback signals by their influence on a quadruped policy, then test policies trained with smaller, task-specific input sets. They use integrated gradients on a simulated A1 and compare key-state policies with full-state and lower-ranked-state policies. In simulated trotting, the key-state policy reaches 99.1% of full-state task performance, averaged across five metrics, not a success rate or real-robot validation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.