This is a simple example of using Unitree Robots for reinforcement learning, including Unitree Go2, H1, H1_2, G1
Install pytorch 1.10 with cuda-11.3:
pip3 install torch==1.10.0+cu113 torchvision==0.11.1+cu113 torchaudio==0.10.0+cu113 -f https://download.pytorch.org/whl/cu113/torch_stable.html
Install Isaac Gym
cd isaacgym/python && pip install -e .
cd examples && python 1080_balls_of_solitude.py
Install rsl_rl (PPO implementation)
cd rsl_rl && git checkout v1.0.2 && pip install -e .
Train:
python legged_gym/scripts/train.py --task=go2
--sim_device=cpu
, --rl_device=cpu
(sim on CPU and rl on GPU is possible).--headless
.v
to stop the rendering. You can then enable it later to check the progress.logs/<experiment_name>/<date_time>_<run_name>/model_<iteration>.pt
. Where <experiment_name>
and <run_name>
are defined in the train config.Play:python legged_gym/scripts/play.py --task=go2
load_run
and checkpoint
in the train config.https://github.com/user-attachments/assets/98395d82-d3f6-4548-b6ee-8edfce70ac3e
https://github.com/user-attachments/assets/a9475a63-ea06-4327-bfa6-6a0f8065fa1c
https://github.com/user-attachments/assets/a937e9c4-fe91-4240-88ea-d83b0160cad5
https://github.com/user-attachments/assets/0b554137-76bc-43f9-97e1-dd704a33d6a9