Reproduce “Test-Time Motion Steering for Perceptive Humanoid Fall Recovery” 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
Osher Azulay, Cheng-Lin Hsieh, Stella X. Yu
RISE steers predictions of a humanoid’s future body states toward task preferences, then generates motion conditioned on those states for fall recovery. On a 29-DoF Unitree G1 in simulation, it achieves 93.0% stable recovery on held-out configurations, versus 80.8% for VIGOR. Real-robot demonstrations include wall-guided route choice, but do not establish broad reliability or guarantee stable recovery.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.