Effective Control of Leader-Follower Networks

The main goal of this project is to develop a theory of input selection for effective control of linear and some nonlinear systems. In contrast to traditional approaches in control theory, which often take both the system and the set of actuated variables as fixed, our starting point will be only the system itself. We will then develop algorithms to decide which variables of the system should be affected with an input to optimize a variety of control objectives. Actuator selection problems appear throughout control and engineering, but the primary motivating application for this work is multi-agent control. It is expected that future military missions will be performed in part by autonomous or semi-autonomous robotic platforms. It is often undesirable to give each drone in a swarm its own commands and an alternative approach is to have a single node be controlled with the remaining nodes using a leader-following method. Algorithm for deciding which variables to actuate can be used to decide which nodes of the swarm should be controlled.