Reproduce “SCRUB: Contact-Based Surface-Frame Estimation for Force-Motion Control with an Armless Quadruped” 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
Sourav Selvaraj, Juan Pablo Espinosa Chessal, Mahdi Agheli
SCRUB estimates a surface normal while a quadruped presses and slides a body-mounted tool across an unmapped surface. A Kalman filter combines force and velocity cues, averaging velocity over a gait cycle to reduce walking-induced ripple, and a predictive controller coordinates the body, feet, and tool reaction. On a Unitree Go1, it completed 5/5 wall traverses and 5/5 column orbits at 12 mm/s and 10 N; results cover two near-vertical surfaces.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.