RATs: Multi-Agent Robot Lifelong Skill Learning Without Gradients or RL
rsasaki0109 · x · 2026-07-30
RATs is a multi-agent Code-as-Policy system for lifelong robot skill learning. During free-form play, LLM agents invent tasks, write code-as-policy, and distill successful executions into a reusable skill library. At evaluation, skills are reused as planner context, with no gradients or RL; all learning through structured natural-language feedback and code reuse.
More from Embodied
- 1X CEO Reiterates Promise: Humanoid Robot Neo to Support Full Autonomy by Year-End Delivery — ChrisGPT · 2026-07-30
- Google Gemini Robotics Set for Major 2.0 Update — CyberRobooo · 2026-07-30
- Google Teases New AI Photography Experiences Coming to Pixel 11 Pro — docmilanfar · 2026-07-30
- Handroid: A Reconfigurable Robot Shifting Between Humanoid and Dexterous Hand — carlosdponx · 2026-07-30
- Prediction Is Not Perception: Why Video Models Fall Short for Embodied AI — PierceLilholt · 2026-07-30
- Satyress Unveils 2-Meter-Tall Centaur Robot for Disaster Zones — seanmcdonaldxyz · 2026-07-30