This simulation demonstrates how AI can learn to drive on a circular (looped) track using genetic algorithms and neural networks. Cars must navigate this randomly generated circuit without any prior knowledge of the environment.
Key Features:
Enhanced Neural Network: Using Sigmoid activation function
Crossover: Combining best traits from top performers
Adaptive Mutation: Rate adjusts as generations progress
Performance Optimization: Delta-time based updates
Improved Collision Detection: Polygon & line-segment based