🐦 Flappy Bird AI using NEAT

An AI agent that learns to play Flappy Bird using the NEAT (NeuroEvolution of Augmenting Topologies) algorithm. This project uses Pygame for game rendering and NEAT-Python for training the neural networks. The agent evolves over generations, improving its ability to navigate pipes without human intervention.

🚀 Features

• 🧠 AI-controlled bird using NEAT
• 🏞️ Real-time graphics using Pygame
• 📈 Generation tracking and scoring
• 💾 Configurable NEAT parameters
• 🔄 Replayable and reproducible simulations

🧰 Technologies Used

• Python 3.x
• Pygame – for rendering and game mechanics
• NEAT-Python – for neuroevolution and fitness evaluation

How NEAT Works in This Project

• Each bird is controlled by a neural network.
• Networks are evolved using NEAT over multiple generations.
• Fitness is based on survival time and number of pipes passed.
• The best-performing networks survive and mutate to improve over time.

Each bird is controlled by a neural network that receives 3 inputs:

• 🐤 Bird’s vertical position
• 📏 Distance to the next pipe (top)
• 📏 Distance to the next pipe (bottom)

The single output decides:
➡️ Jump if the value > 0.5
➡️ Do nothing otherwise

Fitness increases with time survived, and decreases for collisions or bad jumps.

🛠️ Installation

1. Clone the Repository

git clone https://github.com/Tynoee/AI-Flappy-Bird-Game.git