Emergent Collective Decisions in a Swarm of Robots
Trianni Vito, Dorigo Marco

Abstract:

A swarm robotic system is normally characterised by many individuals, each having a partial/limited knowledge about the global pattern of which it constitutes an element. In such a system, decision-making processes may be problematic. However, inspiration can be drawn from insect societies, in which self-organisation plays a crucial role in most of the decisions taken by the colony. In this work, we show how, in a swarm robotic system, a decision can be the result of a collective process: it emerges from the numerous interactions among the individuals and between individuals and environment. We present a task in which a swarm of physically connected, simulated robots has to take a decision whether to pass over a trough or change direction of motion if the gap is too wide to be bridged. We show how such a decision can be collectively taken, based only on a self-organising process.