This project is a clone of the popular Google Chrome "No Internet" Dino game, featuring both an AI-powered mode and a manual player mode. The AI is trained using the NEAT (NeuroEvolution of Augmenting Topologies) algorithm, allowing it to learn and improve its ability to navigate obstacles over multiple generations.
This project serves as an introduction to AI-driven gameplay and evolutionary algorithms, demonstrating how machine learning techniques can be applied to classic games.
Release Date: October 9, 2025
Developer: Simon Oszust
Languages: English & Spanish
- AI Mode – Uses NEAT to train an AI that learns to jump over obstacles and improve its survival time.
- Player Mode – Allows users to play the game manually, replicating the experience of the original Google Dino game.
- Obstacle Avoidance – The AI learns to recognize and react to obstacles such as cacti and birds.
- Neural Network Visualization – Displays the AI's decision-making process in real time.
- The AI evolves over multiple generations, improving its performance based on a fitness scoring system.
- The fitness function rewards survival time and penalizes collisions with obstacles.
- Players can switch between AI mode and manual mode for different gameplay experiences.
- Python
- Pygame for graphics and game mechanics
- NEAT-Python for AI training and evolution
- Storage: 500 MB
- Operating Systems: Windows & MacOS
- Read/write access to files on your device