RL-100 Accepted to Science Robotics with Open-Source Code Release

Excited to share that RL-100 has been accepted to Science Robotics !

We have open-sourced the code here: https://github.com/Lei-Kun/RL-100

Beyond the paper results, RL-100 provides a unified library for diffusion/flow policy RL post-training, iterative offline data flywheels, and real-world robot RL.

A key lesson: real-world robot RL is a system problem. Calibration, rewards, rollout data flywheels, stochastic data collection, randomization, safety checks, and deployment details all matter.

Hope this repo helps make real-world robot RL more reproducible and easier to build on.