Project: NEAT

By Sten

 In this project a self-learning ai will learn how to drive a circuit. It is has an extra focus on getting a fast time. Around the track checkpoints are set-up. When getting a checkpoint the car gets extra live points. These are more live points when reaching the checkpoint faster. NEAT is the evolutionary ai is use. This is a neural network. Only this neural networks only starts with a input nodes and output nodes. With evolving the program adds extra nodes and connections between all the nodes. This makes it so that the program learns how to drive the track. Furthermore, this system has species. When clients look alike enough then they are put in the same species. The program can only breed clients that are in the same species. 

> Read report