An output feedback approach to robust dynamic average consensus
Jemin George, He Bai · 2017
This paper presents the formulation and analysis of an output feedback dynamic average consensus algorithm, where each agent measures an output of a dynamical system subject to an unknown bounded disturbance. The objective is to estimate the average of the states of the dynamical systems based on the local measurements and local interactions between the agents. The proposed algorithm consists of an exponentially stable local observer and a robust dynamic average consensus estimator connected in a cascade fashion. Asymptotic stability of the combined system is theoretically proved and numerically illustrated. The proposed approach is robust to initialization errors and yields zero average-consensus error without assuming known reference signal dynamics or requiring access to the time derivatives of the reference signals. Moreover, the present model assumes heterogeneous local dynamical systems and local bounds on the unknown disturbance-term.