Artificial Life VI


Artificial Evolution of Visually Guided Foraging Behaviour

Ben Hutt
Department of Cybernetics, University of Reading.

Dave Keating
Department of Cybernetics, University of Reading.


Abstract

A population of visually guided agents have been evolved to forage for food pellets using an extended genetic algorithm. Each agent is equipped with an artificial neural network that controls its behaviour and a simple eye which provides input from the environment. The genetic algorithm is responsible for the positioning of light sensitive cells on the retina of the eye and also the design of the neural network. This includes the number, size and sensitivity of the light sensitive cells and also the number of neurons, connectivity and weights contained within the neural network. The system produces complicated and highly recurrent neural network controlled agents that can successfully forage for food pellets.


Back to ALife6 Table of Contents / Keyword Index / Author Index / Meeting Schedule