Graphically train a neural network to avoid obstacles.
A small creature ('annt') is walking around in a square area ('world'). In the world there are obstacles. Goal is to teach the annt to avoid bumping into the obstacles. The direction of the annt's walking is determined by a small neural network. Initially the annt will show random behaviour. It will not avoid the obstacles in view. The user is supposed to train the annt to respond to obstacles in view. Todo: add food to be collected by annt.
Please download files from the project page to further investigate.
Any comments, extensions, etc. are very welcome. Use the tracker on the project page.