Linear Quadratic Leader-Following Consensus of Multi-agent Systems: a Decentralized Computation and Distributed Information Fusion Strategy

This study delves into the leader-following consensus problem in linear multi-agent systems that are structured on weight-balanced digraphs. The primary aim is to formulate a distributed, implementable optimal controller capable of achieving both leader consensus and simultaneous optimization of the linear quadratic cost function. To attain this objective, we introduce a decentralized computation approach and advocate for a distributed information strategy. Initially, the computation of the global Riccati equation is disassembled into the computation of local Riccati equations. Following that, we propose an information fusion algorithm utilizing the dynamic average consensus approach to unravel the optimal controller, enabling its implementation in a distributed fashion. Additionally, we offer numerical simulation examples to demonstrate the efficacy of our proposed approach.