Reproduce “Generalized Inverse Kinematics via Conditional Flow” 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
Kaitlyn Zhang, Tomás Lozano-Pérez, Leslie Pack Kaelbling
GenGenIK learns to generate whole-body configurations and trajectories from end-effector targets, scene geometry, and optional requests. Its masked model samples candidates that are refined numerically, using training examples generated with cuRobo. On simulated RB-Y1 bimanual-lift path tasks, it achieved 92.7% success@32, versus 42.7% for cuRobo, and reported 36.1 ms versus 940.0 ms to first valid solution. Held-out under-table pickup success was 0%.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.