Incremental Perception in Robotic Swarms

Abstract

In biological swarm models, control is decentralised, and whilst a solitary low level-processing agent may fail to achieve a large-scale task, an ensemble of such agents, such as colonies of bees and ants, may succeed. The area of swarm robotics builds on this idea in a way that allows simple robotic agents to perform complex tasks by combining the efforts and capabilities of the individual agents in the colony. In this paper the authors introduce the new concept of incremental perception in the context of swarm robotics. The authors define incremental perception as the ability to combine information perceived by multiple agents so that the swarm as a whole may use it. The authors show how this effect can be simulated and studied, and postulate that it is one of the key requirements that lead to collective behaviour.