I’m working on a simple AI that can walk around inside a maze made of walls ( it doesn’t have to solve it, just to walk around) and I’m trying to figure out how to use a perceptron to teach it to recognize walls around it.
I started by casting several rays around my dude, and collecting the distances between him and the walls around him, then based on the distances I can “weight” up the directions I want to avoid or I want to go to. And that works decent, he can start walking around and when he gets to a curve he “senses” it and turn.
But I would like to update the weights that regulate this behaviour, and I’m not quite sure on what te desired output for each direction should be when I’m inside the weight update function…
can anyone help?