The University of Sussex

Animat navigation using visual landmarks

Peter G.R. de Bourcier

In this paper we describe the results of an initial study into the problem of navigation using visual landmarks. We built a minimal 2-D world inhabited by an Animat, hoping it could learn to mimic various aspects of insect navigation. Empirical results are presented that show basic navigational behaviours based on those found in central place foraging Hymenoptera, can be exhibited in very simple PDP models equipped with rudimentary learning techniques. This paper concludes with our plans for further work in this area.

This paper is not available online