Reproduce “TACOS: Terrain-Aware COntact Supervision for Visual Footstep Planning” 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
Jean-Michel Fortin, William Larrivée-Hardy, Gabriel Jeanson, Olivier Gamache, Daniele De Martini, François Pomerleau, Philippe Giguère
TACOS learns foothold instability from legged robots’ contact experience, converting state-estimator residuals into sparse labels projected onto earlier camera images. A visual predictor trained on hard labels reaches median footstep-level Spearman correlation ρ = 0.554 on unseen outdoor runs; hard labels achieve severity-ranking VUS of 0.893. The field demonstration covers one 50 m route, not broad generalization.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.