This award introduces new theoretical and computational tools for the design of structured optimal controllers for large-scale systems. Standard optimal design methods result in controllers that demand communication among all subsystems. In very large systems, this is a costly scheme due to its extensive communication, computation, and hardware requirements. The PIs will utilize special structures inherent to certain distributed control problems, such as invariances and symmetries, to develop a framework for localized control which compares to that available for centralized design. This research will result in the design of controllers that posses pre-specified localization and communication architectures, and are optimal with respect to global cost functionals. The educational aspect is centered around the development of a new distributed systems curriculum. This will include new introductory courses that will familiarize undergraduate and master-level students with the tools for analysis and design of distributed systems. These courses will emphasize practical applications and physical interpretations.
This award will lead to significant technological advances by enabling the optimal localized control for systems with large number of individual components. Particular attention will be paid to the design of optimal control strategies for large-scale vehicular formations. The PIs will train participating students in a unique interdisciplinary fashion; they will also jointly organize workshops aimed at bringing together a broad audience of students, researchers, and industry professionals. These workshops will enhance a cross-disciplinary character of the proposed work and facilitate transfer of knowledge between the control theory and engineering practice. The outreach component includes the PIs' engagement in public seminars discussing the importance control engineering in transforming the collective decision making strategies employed by natural fliers, runners, and swimmers to engineering applications.
