This research addresses the analysis and design of decentralized feedback control systems. Such systems include: the power grid, wireless networks, and medical tele-robotics. In these applications, distributed information-gathering is taking place but decision-making is often made locally using small subsets of the total information. The emphasis of this research is on controlling decentralized systems in the presence of uncertainty, which can be uncertainty in inter-subsystem communication latency or uncertainty in the dynamics or parameters of the subsystems themselves. The intellectual merits of this project are to advance foundational knowledge in the science of decentralized control and to produce computationally efficient control algorithms. This research has the potential to impact several different application areas and to serve as a guide to practice for engineers and system designers.
Optimal decentralized control is generally intractable, yet certain broad classes of problems can be efficiently solved. This project will leverage recent structural results for these tractable instances with the goal of elucidating universal architectures that allow for simple adaptations in the face of uncertainty in the subsystems or latency in their communication. To achieve this goal, the project will bring to bear tools and insights from the fields of optimal control, robust and nonlinear control, delayed systems, and decentralized control.
