Division of Labour in Simulated Ant Colonies Under Spatial Constraints

Abstract

Division of labour in ant societies is a kind of role model for distributed problem solving. They are so because ants are simple, non-cognitive, distributed and autonomous and yet they solve an optimisation problem that is very complex and dynamic. This is very desirable in computer science, but as of yet not much research has gone into explaining the underlying mechanisms of division of labour. The venue in this paper is to, by means of evolutionary algorithms, find the implications spatial constraints play in the division of labour in a foraging task of virtual ants. The ants differ only in size, and size implies constraints regarding to ease of movement in an environment where obstacles of other ants and clay exists. The results show that spatial constraints do play a major role in the job-task division evolved, in that test setup with increasing constraints exhibited division of labour in that ants of different sizes occupy different spatial areas of the task domain. In the process, we have evolved the behaviour of the ants that underlie the division of labour. This was done via mapping functions and motivation networks.