AgenticRoboRL lifts a 9B robot planner from 18% to 71% on LIBERO-PRO
UCF-Ai-LLMs · hf · 2026-10-22
AgenticRoboRL applies coding-agent post-training to robotics: supervised fine-tuning followed by multi-turn RL across diverse harnesses (VLA skills + control code) trains robot planners without external memory. On LIBERO-PRO, a trained Qwen 3.5 9B planner reaches 71.0% success, up 52.9 points from 18.1%, rivaling frontier proprietary coding models (Claude Code Opus +Mem: 82.4%, Codex +Mem: 72.1%) despite its small size.
More from Embodied
- HKU-led team launches Embodied Playground, 410-task benchmark for model physical intelligence — YiMaTweets · 2026-10-22
- Berkeley and NVIDIA's AW-0 couples code, video, and action streams for robot manipulation — Zhuoyang Liu · 2026-10-22
- Yao Ming takes the stage to talk robotics at a tech event — micoolcho · 2026-10-22
- Matt Wolfe demos 'personal bodyguard robot' standing in a junkyard — Matt Wolfe · 2026-10-22
- Unitree founder says he's betting on recursive self-improvement — micoolcho · 2026-10-22
- Dyna Robotics' humanoid Taku shown in person: 'it works! It's real!' — chris_j_paxton · 2026-10-22