Robot that executes stand-up motion using two servos and reinforcement learning in MuJoCo
source: https://homemadegarbage.com/rl13
3Dmodel/- 3D print models for the physical robotRL/- MuJoCo model, reinforcement learning environment, and trained PPO modelArduino/- M5Atom sketch and exported policy network
Please refer to the README in each folder for detailed usage instructions.
The 3Dmodel/ folder contains STL files for building the physical robot.
footP.stlarm1P.stlarm2P.stlarmhornP.stl
The RL/ folder contains the simulation and trained model.
Key contents:
- MuJoCo model
- Gymnasium environment
- Trained Stable-Baselines3 PPO model
- Playback and evaluation scripts
The Arduino/ folder contains the M5Atom sketch for controlling the physical robot.
Key contents:
robo03.inopolicy_network.h
policy_network.h is a C header file containing the trained policy, used to execute the stand-up motion on the M5Atom.