In our project for the course Programmierpraktikum II we simulate
a multiple agent system. The agents are individuals which "live" in
a two dimensional world. Each agent is controlled by an own
artificial neural network that determines the wandering
direction. The artificial neural network gets its inputs from the
sensors which an agent has. There is one standard direction which
is chosen if no other agents are discovered by the sensors. If
there is an other agent within sensorrange then the artificial
neural network changes the wander direction. The agents live for
one lifecycle, at the end they get a rating. Then the controller
mutates randomly, and a new generation starts to wander. The goal
of this simulation is that the agents learn the best wander behaviour
so that they can visit as many distinct points as possible.