This DRL framework provides a way to train and evaluate policies. The code uses the phsyics engine PyBullet and the Stable Baselines-3 algorithms. Policies can be trained using PPO or TD3, however the code is predominantly structured for PPO implementations as the PPO hyperparameters are defined. The required Python version is Python 3.10. The required pip list is as follows:
pip install pybullet pip install gymmnasium pip install gym==0.21.0 pip install array2gif pip install moviepy pip install stable_baselines3[extra] pip install tensorflow==2.14.0 pip install gin-config pip install protobuf==3.20.3 pip install sympy