Go2 RL GYM

🌎 English | πŸ‡¨πŸ‡³ δΈ­ζ–‡

This repository builds on unitree_rl_gym to train the Unitree Go2 quadruped with reinforcement learning.

|
Isaac Gym
|
Mujoco
|
Physical
| |--- | --- | --- | | ![isaacgym eval](https://raw.githubusercontent.com/robogauge/picture-bed/refs/heads/main/go2_rl_gym/isaacgym_eval.gif) | ![mujoco eval](https://raw.githubusercontent.com/robogauge/picture-bed/refs/heads/main/go2_rl_gym/mujoco_eval.gif) | ![real eval](https://raw.githubusercontent.com/robogauge/picture-bed/refs/heads/main/go2_rl_gym/real_eval.gif) |
## πŸ“¦ Installation Follow the step-by-step setup guide in [setup.md](doc/setup_en.md). ## πŸ› οΈ Usage Guide ### 1. Train Run the following command to launch training: ```bash python legged_gym/scripts/train.py --task=xxx ``` #### βš™οΈ Arguments - `--task`: Required. Options include `go2`, `go2_cts`, `go2_moe_cts`, `go2_moe_ng_cts`, `go2_mcp_cts`, `go2_ac_moe_cts`, `go2_dual_moe_cts`; `go2_moe_cts` is the paper's final version. - `--headless`: Render viewer by default; set to `true` to disable rendering for higher throughput. - `--resume`: Resume training from a chosen checkpoint in the logs. - `--experiment_name`: Experiment folder to save/load from. - `--run_name`: Run subfolder name to save/load from. - `--load_run`: Name of the run to load (defaults to the most recent run). - `--checkpoint`: Checkpoint index to load (defaults to the latest file). - `--num_envs`: Number of parallel simulated environments. - `--seed`: Random seed. - `--max_iterations`: Maximum training iterations. - `--sim_device`: Physics simulation device. Use `--sim_device=cpu` to force CPU. - `--rl_device`: RL computation device. Use `--rl_device=cpu` to force CPU. - `--robogauge`: Enable RoboGauge evaluation tool; disabled by default. Evaluation results are saved as `results_{it}.yaml` in `logs/{exp_name}/{date}/robogauge_results` and logged to TensorBoard. - `--robogauge_port`: RoboGauge server port; default is 9973. > RoboGauge evaluation requires a separate server to be started. Refer to the [RoboGauge documentation](https://github.com/wty-yy/RoboGauge). **Default checkpoint path**: `logs//_/model_.pt` --- ### 2. Play Visualize policies inside Gym with: ```bash python legged_gym/scripts/play.py --task=xxx ``` **Notes** - Play launches on randomized terrain with difficulty between 7 and 9. - It automatically loads the latest checkpoint inside the experiment folder. - Override via `experiment_name` and `checkpoint`, for example: ```bash python legged_gym/scripts/play.py --task=go2_cts --num_envs 100 --experiment_name go2_cts_hard_terrain --checkpoint 100000 ``` #### πŸ’Ύ Policy Export Play exports the Actor network to `logs/{experiment_name}/exported/policies`: - `policy.pt`: TorchScript model for Sim2Sim. - `policy.onnx`: ONNX model for Sim2Real. - `policy.pkl`: Raw weights. #### Demonstration ![isaacgym play](https://raw.githubusercontent.com/robogauge/picture-bed/refs/heads/main/go2_rl_gym/isaacgym_play.gif) --- ### 3. Sim2Sim (Mujoco) Run policies in the Mujoco simulator: ```bash python deploy/deploy_mujoco/deploy_go2.py ``` Connect an Xbox-compatible gamepad to enable teleoperation; otherwise, the agent keeps a default forward command. - **Swap the policy**: The default checkpoint is `deploy/pre_train/go2/go2_cts_150k.pt`. Replace `policy_path` in the YAML config with your own `logs/{experiment_name}/exported/policies/policy.pt`. - **Swap terrains**: Default terrain is `resources/robots/go2/stairs.xml`. Alternatives include `flat.xml`, `race_track.xml`, `cross_stairs.xml`, and `cross_slope.xml`. Generate new terrains with [windigal - mujoco_terrains](https://github.com/windigal/mujoco_terrains) #### Results | Flat | Stairs | Race Track | |--- | --- | --- | | | | | --- ### 4. Sim2Real #### 4.1 Python Deployment (requires [unitree_sdk2_python](https://github.com/unitreerobotics/unitree_sdk2_python)) ```bash # Onboard Jetson: pick Python by JetPack version # JetPack 6: Python 3.10 # JetPack 5: Python 3.8 conda create -n deploy python=3.10 conda activate deploy # Install the matching PyTorch wheel for your Jetson # https://forums.developer.nvidia.com/t/pytorch-for-jetson/72048 git clone https://github.com/unitreerobotics/unitree_sdk2_python.git cd unitree_sdk2_python pip3 install -e . ``` In the Unitree app, open Device β†’ Service, disable `mcf/*`, and enable the `ota_box` service. Assuming the interface to the low-level controller is `eth0`: ```bash cd deploy/deploy_real python deploy_real_go2.py eth0 ``` Press `start` to stand and `A` to engage the controller. #### 4.2 C++ Deployment (requires unitree_cpp_deploy) Follow the usage described in [unitree_cpp_deploy](https://github.com/wty-yy-mini/unitree_cpp_deploy). #### Demonstration | Python Deploy | C++ Deploy | | --- | --- | | ![python deploy](https://raw.githubusercontent.com/robogauge/picture-bed/refs/heads/main/deploy/py_deploy_with_commands.gif) | ![cpp deploy](https://raw.githubusercontent.com/robogauge/picture-bed/refs/heads/main/deploy/cpp_deploy_with_commands.gif) | --- ## πŸŽ‰ Acknowledgements This repository would not exist without the following open-source projects: - [unitree_rl_gym](https://github.com/unitreerobotics/unitree_rl_gym): Unitree's core RL training framework. - [legged_gym](https://github.com/leggedrobotics/legged_gym): Base locomotion environment. - [rsl_rl](https://github.com/leggedrobotics/rsl_rl.git): Reinforcement learning algorithms. - [mujoco](https://github.com/google-deepmind/mujoco.git): High-performance CPU physics simulator. - [unitree_sdk2_python](https://github.com/unitreerobotics/unitree_sdk2_python.git): Python hardware interface for deployment. - [unitree_sdk2](https://github.com/unitreerobotics/unitree_sdk2): C++ hardware interface for deployment. Related publications implemented in this repo: - [CTS: Concurrent Teacher-Student Reinforcement Learning for Legged Locomotion](https://arxiv.org/pdf/2405.10830) Contributors: - [@windigal](https://github.com/windigal): CTS algorithm reproduction, terrain generation, video editing - [@wertyuilife2](https://github.com/wertyuilife2): CTS algorithm reproduction --- ## πŸ”– License New contributions follow the [MIT License](LICENSE); the original unitree_rl_gym remains under the [BSD 3-Clause License](LICENSE). See the complete [LICENSE file](LICENSE) for details.