Evolving controllers for programmable robots to influence non-programmable lifeforms: A casy study

Abstract

In this paper, a decentralized reaction-diffusion-based controller is evolved for a set of robots in an arena interacting with two simulated juvenile bees as non-programmable agents. The bees react to the stimuli that are emitted by the robots. The evolutionary process successfully finds controllers that produce proper patterns which guide the bees towards a number of given targets. The results show a preference of heat as the dominant stimulus causing movement of the bees.