Neuroevolution-based Driving AI
View this project on GitHub: https://github.com/EthanGahm/neuroevolution-driving-aiVideo Overview
Project Description
A basic, neuroevolution-based driving simulator written in Processing.
The algorithm employs a fixed-size neural network with genetically-inherited weights. Indivuals attempt to navigate a track while avoiding walls. Fitness is determined based on the number of invisible checkpoints reached along the length of the track. After all indivduals of a given generation have collided with a wall, the next generation begins, with the most successful individuals passing down their traits. Within only a handful of generations, most individuals learn to navigate the entire length of the track.
Technologies Used
- Processing 3 - All code