Concurrent learning-based network synchronization

A data-driven concurrent learning-based control law is developed for the synchronization of a leader-follower network of agents with uncertain nonlinear dynamics wherein only a subset of the follower agents is connected to the leader. The development is facilitated by the use of online data-driven adaptive update policies to approximately learn a distributed control law which satisfies a given performance metric without the need for persistence of excitation (PE). A neighbor-decoupled control structure is introduced which provides greater flexibility in the consideration of individual neighbors during synchronization and makes the control of each agent a differential game.