Decentralized flocking protocol of multi-agent systems with predictive mechanisms

This paper presents a novel decentralized predictive flocking protocol of discrete-time multi-agent systems, where every agent is driven by a predictive controller with the local information of its neighbors. The feasibility of the decentralized protocol and system convergence is demonstrated by numerical simulations, where the effects of sampling period, prediction horizon and control horizon, the three parameters for predictive controllers, on the flocking performance are verified with the supportive illustrations of convergence rate and lattice regularity.