Collective search and decision-making for target localization

Abstract

In this article we investigate the properties of collective search and decision-making in robotic swarm, inspired by a phenomena witnessed in bio-societies. The task of the proposed robotic swarm, comprising scouts and labourers, is to find the most hazardous target in a predefined area. Since in the proposed scenario the time interval for decision-making is predefined, robotic scouts have to detect targets within a particular amount of time. Hence, in the first part of the article, we define a model of scout movement that enhances the explored area. As we want to keep the searching process as simple as possible, and at the same time to mimic social insect behaviour, a particular type of correlated random walk is used for exploration. The second part of the article deals with modelling of the decision-making process in the robotic swarm. Using random walk theory we determine under which circumstances all agents (or a particular number of them) would be committed to the most hazardous target at the moment when the predefined time interval for decision-making expires.