README.md

Lunar-Lander-Environment

Code and relevant files for the final project of CM50270 (Reinforcement Learning) for MSc. in Data Science at University of Bath.

This is an implementation of DQN, DDQN, DDPG and TD3 on Lunar Lander environment from OpenAI Gym.

Notes for viewer:

1. DQN and DDQN:

   a. Are implemented on Discrete action space.

   b. Use "LunarLander-v2"

   c. RANDOM_SEED = 42

2. DDPG and TD#:

   a. Are implemented for the continuous action space.

   b. Use "LunarLanderContinuous-v2"

   c. RANDOM_SEED = 101

Solving Issues related to Box-2d:

    pip uninstall pygame
     pip install gym[box2d] pygame

Still Facing Issues?

Try:

    pip uninstall pygame
    pip install swig
    pip install pygame

For Google Colab:

1. Use these two cells at the begining of the Colab file

   [1]:

    !apt-get update
   
    !apt-get install -y build-essential swig
   

   [2]:

    !pip uninstall -y box2d-py
    
    !pip install -U --no-cache-dir --force-reinstall box2d-p
   

Youtube Link for Agent Training Video: https://youtu.be/7UMosL_k5Vc Youtube Link for Presentation: https://youtu.be/NSDx5YWXJYY