AI-learns-to-play-Flappy-Birds

Here is the source code for a Python project that implements a machine learning algorithm in the Flappy Bird video game using neural networks and a genetic algorithm.

All code is written in Python using Pygame and NEAT for genetic algorithm implementation.

Neural Network Architecture

To play the game, each unit (bird) has its own neural network consisted of 2 layers:

1. an input layer with 3 neurons presenting what a bird sees:

   1) bird's height
   1) height difference between the bird and the next top Pipe
   2) height difference between the bird and the next bottom Pipe
   

2. an output layer with 1 neuron used to provide an action as follows:

   if output > 0.5 then flap else do nothing
   

The Main Concept of Machine Learning

The main concept of machine learning implemented in this program is based on the neuro-evolution form. It uses evolutionary algorithms such as a genetic algorithm to train artificial neural networks. Here are the main steps:

1. create a new population of 50 units (birds) with a random neural network

2. let all units play the game simultaneously by using their own neural networks

3. for each unit calculate its fitness function to measure its quality as:

   fitness = 0.1 * total travelled distance + 5 * number of pillars passed through.
   

4. when all units are killed, evaluate the current population to the next one using genetic algorithm operators:

   1) selection
   2) crossover
   3) mutation
   

5. go back to the step 2

Implementation

flappyBirds.py

The entire game logic is implemented in flappyBirds.py file.

Base.py

Class to represent the moving floor.

Bird.py

class to represent a Bird sprite.

Pipe.py

Class to represent a moving barrier. This contains a top and bottom sprite.

config-feedforward.txt

configuration file to modify neuroevolution algorithm parameters.