project_resource

Half-Cheetah
- Class https://github.com/openai/gym/blob/master/gym/envs/mujoco/half_cheetah.py
- MuJoCo XML https://github.com/openai/gym/blob/master/gym/envs/mujoco/assets/half_cheetah.xml

Important Paths
-SB3
/home/jpc/anaconda3/envs/mujoco210_openai_gym/lib/python3.9/site-packages/stable_baselines3
-Gym MuJoCo Envs
/home/jpc/anaconda3/envs/mujoco210_openai_gym/lib/python3.9/site-packages/gym/envs/mujoco
- Gym MuJoCo XML Assets
/home/jpc/anaconda3/envs/mujoco210_openai_gym/lib/python3.9/site-packages/gym/envs/mujoco/assets/
- mujoco_py
/home/jpc/anaconda3/envs/mujoco210_openai_gym/lib/python3.9/site-packages/mujoco_py
- registry of environments (can add my custom envs too)
/home/jpc/anaconda3/envs/mujoco210_openai_gym/lib/python3.9/site-packages/gym/envs/__init__.py

"half_cheetah_v#" contains init of XML model and defines observations, reward function, reset init, etc
"half_cheetah.xml" contains MuJoCo model of cheetah, scene, environment, cameras, etc

Already trained RL agents
https://github.com/DLR-RM/rl-baselines3-zoo#enjoy-a-trained-agent
(MuJoCo compatibility) https://github.com/DLR-RM/rl-baselines3-zoo#mujoco-environments
** How to use pre-trained agent in python script?
Pre-trained algs for each env can be found in rl-baselines3-zoo/rl-trained-agents - then from the proper env folder and
alg folder, you must take the .zip file of the model and move it to the directory of your script. Then you can use all
the load/save/train command with it
    - saving/loading models https://stable-baselines3.readthedocs.io/en/master/guide/save_format.html#
    - ex train/save/load https://stable-baselines3.readthedocs.io/en/master/guide/examples.html#basic-usage-training-saving-loading

How to create custom gym env - guide
https://github.com/openai/gym/blob/master/docs/creating_environments.md
https://stable-baselines3.readthedocs.io/en/master/guide/custom_env.html
https://colab.research.google.com/github/araffin/rl-tutorial-jnrr19/blob/master/5_custom_gym_env.ipynb
(challenge is creating pip package and installing)

** How to modify local gym repo to create custom environments
1) copy & paste Python environment file (ex. half_cheetah.py) in envs/mujoco and corresponding model XML file in
 envs/mujoco/assets, and rename (if making changes to 1 or both)
2) update registry in envs/__init__.py with details of new env
 [rename of register ID must follow reg exp in registry.py]