Here is the source code for RL trainning for Talos robot.
This project are build under Python 3.63 and Pybullet environment. Tested on Windows 10 and Linux (testing).
The following instruction would lead you to created an environment to run this project succesfully.

The robot are aim to raise his left arm during the trainning in this vision.
The basic function of Talos are tested.

Learning to Stand.
Talos's joints was not returned to origin position sometimes after each episode (Obsevered by GUI.)
Issues about negative-log in tensorboard.

PID controlled.
Identidy the Controlled Joints.
Basic environment set-up and tested.

Here is related API and packages use in this project.

Python environment <=3.70 are required to run this project.
Python 3.6.13 are used to build the whole project.

To install Pybullet use following code in anaconda(Windows) and command line (Linux):
pip install pybullet
For more information on PyBullet, read PyBullet .

To install Gym use following code in anaconda(Windows) and command line (Linux):
pip install gym
For more information on PyBullet, read Gym offical installation documentation.

To install Torch use following code in anaconda(Windows) and command line (Linux):
pip install torch
For more information in Pytorch, read Pytorch offical cite.

To install example_robot_data use following code in anaconda(Windows) :
conda install example-robot-data -c conda-forge
For Linux Users: Please following instruction on Gepetto's page.

To install Torch use following code in anaconda(Windows) and command line (Linux):
pip install stable-baselines
For more information in Stable-baselines, read Stable-Baseline.

Notification: Tensorflow >= 2.0 are not support for stable-baselines. This project are using TF 1.15.0 for build.
Python >= 3.8 require Tensorflow 2.2 or higher to run
To install Tensorflow use following code in anaconda(Windows):
conda install tensorflow==1.15
Required Python 3.6 or 3.7
To install Tensorflow on Linux:
pip install tensorflow==1.15
If you need more helps on Tensorflow, Please check Pypi for Tensorflow and Tensorflow.