Robust Data-Driven Containment of Fully Unknown Heterogeneous MASs From Noisy Data

Abstract

This article investigates the robust data-driven containment control problem for heterogeneous multiagent systems where the system dynamics of both the leaders and the followers are unknown. During the data collection phase, the follower systems are disturbed by unmeasurable but bounded noises. A data-based linear matrix inequality condition is first constructed from the noisy data to determinate the feasible feedback gain for each follower. Then, the distributed observer which is independent of the system matrix of the leaders is designed to estimate the convex hull of the leaders. Moreover, the approximate solution to the linear matrix equation for heterogeneous system is solved with bounded approximate error where the noisy data of each follower and the normal data of arbitrary leader is utilized. Based on the proposed feedback gain and observer as well as approximate solution, the robust data-driven control protocol is provided to guarantee the uniform boundedness of containment error. Finally, a numerical example and a multivehicle model are given to verify the effectiveness of the designed containment control protocol.