Snake Simulation

A Python snake game where an AI agent learns to play entirely on its own using a genetic algorithm. Each generation spawns a population of snakes with randomised neural network weights. The snakes that survive longest and eat the most food reproduce and mutate, slowly evolving effective play strategies.

The simulation is visualised in real time with Pygame: you can watch each generation get progressively better, from random thrashing to deliberate food-seeking behaviour, usually within 30-50 generations.

The project was built for my Artificial Intelligence course and deepened my understanding of evolutionary computation, fitness function design, and the surprising effectiveness of simple selection pressure.