36
This project implements a self-learning Flappy Bird game using neural networks and a genetic algorithm. The AI learns to navigate through pipes by evolving better strategies over generations. Each bird is controlled by a neural network that decides when to flap based on environmental inputs. The population evolves over generations, with the best performers passing their traits to the next generation.
Real-time learning visualization
Interactive neural network display
Generation statistics tracking
Smooth gameplay animation
Automatic evolution process
Modern, clean UI
Neural Network Architecture: Input Layer with 4 neurons, 2 Hidden Layers with 3 neurons each, and an Output Layer with 1 neuron
Genetic Algorithm: Population Size of 1000 birds, Top 10 performers chosen as parents, Mutation Rate of 5% per weight
Game Parameters: Constant horizontal speed, Gravity of 0.0003, Flap Strength of -0.008 velocity, Pipe Generation every 1500ms
No additional dependencies or setup required. Works in any modern browser with HTML5 canvas support.
Built with