This was created for my Topics in NeuroEvolution and Artificial Embryogeny (CAP 6938)
class headed by Ken Stanley
. For my class project I decided to do a space sim where the spaceships learned to follow a target. My thinking is that this could be extended to control AI in a game, or maybe even robots in real life.
What you'll see in the video is the spaceships learning, over the course of only 30-45 seconds, how to follow the target. They start out with no knowledge of how to even move and by the end they have no trouble keeping up. Run 2 is the better one, so watch that, if anything :)
The project homepage is located at hiranipra.com