• Simple Q-Learning
• Q-Learning (with ε-greedy policy)
• Deep Q-Network

This set of slides present our work in a concise way. You might want to check it out.

A more detailed account about this project have been documented on this blogpost here.

• Python 3
• pygame
• scikit-image
• Keras 2

python flappy_rl.py Run

python flappy_rl.py Train

To train a fresh model, delete the file qvalues.txt before executing the above command.

python flappy_rl.py Run greedy

python flappy_rl.py Train greedy

To train a fresh model, delete the file qvalues_greedy.txt and then execute the above command.

python dqn.py Run

python dqn.py Train

To train a fresh model, delete the file dqn.h5 and then execute the above command.

https://github.com/sourabhv/FlapPyBird