Two-Layer Distributed Protocols for Cooperative Tracking in Mixed-Order Interconnected Multiagent Systems

Using relative measurements in agents' neighborhoods, graph theoretic distributed algorithms have been widely investigated to achieve agreement in a group of individual dynamical systems. We consider a class of partially known multiagent systems with mixed first- and second-order agents whose nonlinear dynamics interact over two agent layers. We assume a constant reference command is sent to only a few agents and propose a cooperative reference tracking problem from a large-scale system viewpoint. We investigate the case that no local control implementation is permitted and, proposing a linear distributed protocol with two communication topologies as control layers, formulate the problem as two graph topology design tasks to be solved using an optimal control theoretic approach. We develop closed-form solutions for these control layers and prove the robustness of exponential reference tracking behavior with respect to the modeling uncertainties over agent layers. We also investigate the robust performance of closed-loop multiagent system based on a quadratic cost function. We verify the feasibility of these developments through a simulation study.