CoRL paper GLIDE uses LLM-edited teleop code, lifting hard-task success from 0% to 90%
anand_bhattad · x · 2026-11-09
Robot imitation learning relies on human teleop demos, but some tasks are too hard to demonstrate — raw VR teleop success rates are 0–10% for tasks like picking cherry tomatoes or pouring wine. The CoRL paper GLIDE uses LLMs to smartly edit teleop code, boosting success to 90%. The authors pitch it as a refreshingly robotics-first way to use LLMs.
More from Embodied
- Stanford team teaches an AI agent a pool table's physics by letting it take real shots (CoRL) — drmapavone · 2026-11-09
- Furong Huang Lab brings six human-video-to-robot works to CoRL 2026 — furongh · 2026-11-09
- Three CMU robot papers spotlighted at CoRL 2026: memory, vision-touch steering, world models — pabbeel · 2026-11-09
- CoRL 2026 balloons to 5,000+ attendees; Princeton PRISM brings 6 robotics papers including oral MolmoB0T — shahdhruv_ · 2026-11-09
- UW RAIVN Lab brings GAM, MolmoAct2 and Reflex to CoRL: 3D grounding beats VLA baselines — LINJIEFUN · 2026-11-09
- BitRobot Ecosystem Lightning Talks Day 1: MolmoAct2, RoboEval, RoboTTS Lineup — DJiafei · 2026-11-09