tarunk04 / flappy_bird_reinforcement_learning

Reinforcement learning implementation to train AI to play flappy bird game

Flappy_Bird_Reinforcement_Learning

This is a simple Reinforcement Learning application on Flappy bird game.

Flappy Bird

Flappy Bird game used in this project is used from repository: https://github.com/sourabhv/FlappyBirdClone

Installation Guide

1. Python 3.5.x and PyGame 1.9.X is required

2. Clone this repository: https://github.com/tarunk04/flappy_bird_reinforcement_learning.

3. Run python train.py.

4. Wait for training to be done (Models will be saved in the folder Models of the repository).

5. Run python flappy.py enter location of trained model and press Enter to see the performance of the model.

NOTE Trained model is already provided in the folder Models/Samples/flappy_bird_model_final.h5. Run python flappy.py and enter anything as text to load default path and enjoy...

ScreenShot

High Score

High Score so far achieved using the trained model ( provided in models/samples/) is : 25378

---