COIL: self-supervised robot imitation learning with 3D keypoint trajectories debuts at IROS 2026
YuXiang_IRVL · x · 2026-09-27
Researcher QuantummCookie presented COIL (Correspondence-Oriented Imitation Learning) at two IROS 2026 workshops (CompRobotics and WORLDS).
- COIL is a self-supervised framework for robotic manipulation
- Core idea: uses 3D keypoint trajectories as a flexible, unified task representation
Related event: COIL: Self-Supervised Robot Imitation Learning via 3D Keypoint Trajectories(2 posts)→
More from Embodied
- Founders Inc Is Building a Personal Flying Car, First Prototype Coming Soon — IsaiahBallah · 2026-09-27
- Reimagine Robotics has customers teach factory robots, cutting per-skill training from 1 day to 10 minutes — CyberRobooo · 2026-09-27
- Four NeurIPS 2026 papers: code as action interface boosts spatial reasoning by 13.6 — CMHungSteven · 2026-09-27
- Boston Dynamics' electric Atlas pivots from backflips to factory work with 56 DOF and tactile sensing — Olivier__OG · 2026-09-27
- Why household robots and home-defense robots won't stay separate for long — facontidavide · 2026-09-27
- OpenAI is hiring SLAM engineers — start with MIT's free 'SLAM for Dummies' guide — lukas_m_ziegler · 2026-09-27