NAO-V40 pybullets training environment

Description

This is NAO-V40 training environment based OpenAI Gym standards. Pybullets is used as the simulationg environment.

Repository structure

├── baselines
│   ├── nao.py
│   └── test_walk.py
├── HumanoidRL
│   ├── envs
│   │   ├── humanoidRL.py
│   │   ├── __init__.py
│   │   ├── test_script.py
│   │   ├── Utility.py
│   │   └── walk_positions.pckl
│   ├── humanoid
│   │   ├── LICENSE.txt
│   │   ├── meshes
│   │   └── nao.urdf
│   └── __init__.py
├── README.md
├── setup.py
├── spinup_run.py
└── test_policy.py

• baselines contains the reference items.
• envs contains humanoidRL script which it the standard Gym environment. Utility class contains the basic functions for communicating with the simulator.
• humanoid contains the meshes and basic physics configrations for NAO V40.

Installation of Humanoid Environment

• Ensure that you are using Python3.

git clone https://github.com/taapasX28/HumanoidRL
cd HumanoidRL/
pip install -e .

Testing

• test_script.py is the testing script for utlity functions and the environment seperately.
• Use python test_script --env=True to test environment.
• Use python test_script --util=True to test utilities.

Running training

• Run spinup_run.py. You need spinningupai installed for that.

Loading models

• Run python test_result.py --file <path_to_directory> to load and run the saved model.
• BUG: Change the feed_dict input from x[None,:] to x[None,:][0].T in spinup.utils