Hierarchical Decentralized Controller Synthesis for Heterogeneous Multi-Agent Dynamical Systems by LQR

Abstract

This paper is concerned with systematic ways of designing hierarchical decentralized controllers for heterogeneous multi-agent dynamical systems. Given a bunch of independent agents or subsystems with a class of information networks, the aim of the paper is to propose a systematic design procedure for hierarchical decentralized controllers, where each subsystem cooperatively interacts with each other as well as controls it locally to achieve both the local and global goals in some senses. It is shown that employing the LQR (Linear Quadratic Regulator) method with properly chosen weighting matrices in the performance index, both the local and global objectives can be achieved by the desired hierarchical decentralized structure which fits the given information network. The effectiveness of the proposed design method is confirmed through an illustrative example and its application to a velocity consensus problem in vehicle platoons.