Evolving control for preserving connectivity among agents of network with non-cooperative moving agents

This paper considers the problem of maintaining connectivity in a so-called noncooperative network of agents. The agents consist of two classes: one class of agents, which is termed uncontrolled, performs some kind of surveillance or monitoring operation, and the trajectory of the agents within this class cannot be altered. The second class of agents, which are termed controlled agents, are used to maintain connectivity. The paper considers the development of a high level mission planner which develops the control policies for the controlled agents. Sliding mode observers are employed to predict the trajectories of the uncontrolled agents and this information is used to construct state dependent graphs encapsulating the measure of connectivity. A form of model predictive control is then employed to develop the control policies for the controlled agents to maximize connectivity.