COLLECTIVE PERCEPTION IN A SWARM OF AUTONOMOUS ROBOTS

Abstract

We present a study that aims at understanding how perception can be the result of a collective, self-organising process. A group of robots is placed in an environment characterised by black spots painted on the ground. The density of the spots can vary from trial to trial, and robots have to collectively encode such density into a coherent flashing activity. Overall, robots should prove capable of perceiving the global density by exploiting only local information and robot-robot interactions. We show how we can synthesise individual controllers that allow collective perception by exploiting evolutionary robotics techniques. This work is a first attempt to study cognitive abilities such as perception, decision-making, or attention in a synthetic setup as result of a collective, self-organising process.