The Robotic Swarm Contamination Problem

Abstract

The problem of reaching a consensus in a swarm of robots has been widely examined in the literature. In this paper we introduce the contamination problem, a new version of the consensus problem in which the robots' influence to adopt a certain value towards a consensus is based on both external factors (its surroundings) and its internal state. The problem's innovation is in the combination of two aspects: first, the existence of adversarial forces that may intentionally act to divert the robots from a desired consensus, and an asymmetry of state transition, that is, a robot may have a stronger tendency to change from one state to another, compared to the other way around. We provide a theoretical analysis of formation creation as a mean to guarantee consensus, and an empirical evaluation-both in simulation and in real robots-demonstrating the influence of different robot behaviors and internal factors on the convergence time and value.