DECENTRALIZED CONTROL OF NETWORKS OF UNCERTAIN DYNAMICAL SYSTEMS By JUSTIN RICHARD KLOTZ A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy DECENTRALIZED CONTROL OF NETWORKS OF UNCERTAIN DYNAMICAL SYSTEMS By Justin Richard Klotz May 2015 Chair: Warren E. Dixon Major: Mechanical Engineering Multi-agent networks, such as teams of robotic systems, benefit from the ability to interact and sense the environment in a collaborative manner. Networks containing agents operating under decentralized control policies, wherein only information from neighboring agents is used to internally make decisions, benefit from autonomy: each agent is encoded with a local objective and has no need to maintain contact with a network coordinator. Such an interaction structure reduces the communication bandwidth and the associated computational requirements in comparison to centralized control schemes, wherein a single network coordinator computes a control policy for each agent based on communicated information. However, the development of decentralized control policies should address the deleterious effects accompanied by the decentralized network structure, such as cascading effects caused by exogenous disturbances, model uncertainty, communication delays, and reduced situational awareness. Chapter 1 motivates the current challenges in the field of decentralized control, provides a comprehensive review of relevant literature, and discusses the contributions of this dissertation. Chapter 2 details the development of a novel, decentralized control policy which provides asymptotic convergence of the states of a leader-follower network of autonomous agents despite the effects of uncertain nonlinear dynamics and unknown