Reproduce “Robot-WATCHDOG : failure detection through object-centric graph representation” 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
Quentin Rolland, Fabrice Mayran de Chamisso, Jean-Baptiste Mouret
Robot-WATCHDOG detects manipulation failures by tracking objects, their motion, and pairwise spatial relationships over time. It represents object centroids and the end-effector in a graph, then scores deviations using trajectory projection and correlation or future-state prediction, with thresholds calibrated on nominal episodes. On BotFails, TPC reaches AUPR 0.811 versus 0.442 for AP-RMS, but evaluation excludes deformable and fluid tasks.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.