Multi-agent systems control over complex networks

Abstract

Multi-agent systems refer to systems with many interconnected individuals, such as a large group of robots or vehicles. Controlling multi-agent systems mimics some natural phenomena of flying birds, swimming fish, and migrating insects, where collective behavior such as motions may emerge from sub-groups of individuals through limited and local interactions among them. This project aims at studying a random multi-agent system with a predefined, independent leader. The leader plays the role as a controller of the global behavior of the whole multi-agent system and each of the other member agents updates its state based on the current information available from its neighboring agents. After that, a computer model is built using Java IDE to monitor the movements of the whole multi-agent system. Real cases are also studied. Finally, how to control the systems to achieve desired configurations and further improvements are discussed.