Reinforcement Learning : Flappy Bird Text

The notebook presents the implementation of the Q-Learning and Monte Carlo algorithms to solve the Text-Flappy-Bird game

Text flappy bird game

The implementation of the environment can be found here: https://gitlab-research.centralesupelec.fr/stergios.christodoulidis/text-flappy-bird-gym

Performance

The average score achieved by both agents on a given episode (during testing):

Q-Learning: 127
Monte Carlo: 100,000

Note: The Monte Carlo agent has perfectly master the game (no death of the bird) after 4000 episodes of training. Thus the episode score has been limited artificially and correspond to the score limit I fixed at 100,000.