evo-rl

Open-source offline reinforcement learning pipeline for SO101 and PiPER robots, providing data collection, training, and evaluation in a closed loop for robotics researchers.