Control of Multiagent Systems with Local and Global Objectives: Experimental Results

Abstract

We focus on multiagent systems that have capabilities to accomplish global and local objectives. The former refers to a task (e.g., reaching a consensus, following a leader and so on), which is expected to be performed eventually by all agents in the system, and the latter is a kind of the secondary objective of the multiagent system, which is desired to be achived by only a certain number of agents without deteriorating the global objective of the multiagent system. To this end, the authors of [1] proposed five different distributed protocols having comparable advantages with system-theoretic stability analyses for multiagent systems being subject to a global objective, which is a leaderless consensus, and local objectives. In addition, we slightly modify these protocols in order to achieve the leader-follower consensus as a global objective. The contribution of this paper is to conduct experiments on a team of ground mobile robots with the protocols mentioned above and to make comparisons of their performances.