Minimalistic Control Scheme for the Development of Search Tasks with Flocks of Robots

Abstract

Abstract This paper proposes a minimalist strategy of agent coordination in a multi-robot system with flocking behavior intended for the development of search tasks in dynamic and unknown environments. The minimalist design principle seeks the ability to implement the strategy on small low-cost robots, with hardware limitations, and with small functional variations among them, a real case of most robotic platforms. We also seek a robust decentralized strategy (without central control) where all agents are virtually identical from the functional point of view, and therefore the damage of a fraction of these agents does not prevent the development of the task. In this sense, our scheme is based on local readings, from which the agents identify the region of the environment in which they find themselves, including obstacles and neighboring agents, and from this information autonomously establish their movement strategy to produce the flocking dynamics to the system. The navigation of the environment is guaranteed by relaxed ergodic movement rules, and the convergence of the search process is achieved by grouping a given number of agents in a particular region of the environment. The strategy was successfully evaluated by simulation by replicating the functional characteristics of real robots and scaling them to large populations.