We apply DQN algorithm to make and artificial agent learn how to land space-craft on moon. The code is explained in the Deep_Q_network.ipynb. You guys are welcome to imporve the hyperparameters or even the algorithm for better performance(check step 5:Explore of the notebook). I have also provided a PDF_research_paper to explain Deep Q-Networks. This project is based on this research paper.
https://gym.openai.com/envs/LunarLander-v2/
Simply execute this on your shell: $pip install -r requirements.txt
Note: The user must install pytorch according to the specifications on his/her workspace. I used torch 1.4.0 and torchvision 0.4.2.
You can skip running the dqn method to avoid train and just run the code using pretrained weights in checkpoint.pth i have provided.
