A.I. Learns to Avoid Obstacles

Checkout my website for more details: http://brainyloop.com/ I created an Artificial Intelligence program that teaches itself to avoid obstacles and navigate in a virtual environment using a machine learning technique called NEAT (Neuro-Evolution of Augmenting Topology). Each generation a new population of agents are created based on the best fit agents from the previous generation. The fitness of agents is measured based on the number of cells they visit each round. The graphs on the right side show the fitness and the brain (Neural Network) of the best fit agent from last round. The agents are given 5 sensors for detecting collision but they have to calibrate the weight of each sensor and the effect they have on their movement. Over time each generation optimize their sensors and their reaction. Please subscribe and comment if you like more content like this. Music Credit: –––––––––––––––––––––––––––––– Dayspring by Firefl!es https://soundcloud.com/mangesh-dongar...​ Creative Commons — Attribution 3.0 Unported — CC BY 3.0 Free Download / Stream: https://bit.ly/_dayspring​ Music promoted by Audio Library https://youtu.be/Zg_P1JQV5pY​ ––––––––––––––––––––––––––––––