On the Control of Multi-Agent Systems: A Survey

Recent years have witnessed a trend to use networked multiple autonomous agents to accomplish complex tasks arising from space-based applications, smart grids, and machine learning. Such systems are referred to as multi-agent systems, where a global objective is achieved via the local interactions among the agents. The recent two decades have witnessed a rapid development of MASs in automatic control, but their root can be traced back much earlier. This paper reviews the research progress of MASs in past years. After briefly introducing the basic concepts and definitions, we discuss agents’ dynamic models, including both linear and nonlinear models, describe different cooperating tasks, such as consensus, coordinating tracking, formation control, distributed average tracking, distributed estimation, containment control, surrounding control, and distributed optimization. We introduce various research issues for MASs, including time-delays, noise or disturbance, quantization, connectivity maintenance, and event-triggered control. We present different MAS control algorithms, including proportional control, proportionalintegral control, adaptive control, model predictive control, passivity-based control, and nonsmooth control, and their applications in multi-robot systems, sensor networks, smart grid, machine learning, social networks, and many-core microprocessors.