Control and Sensing under Limited Information

University Of Illinois At Urbana-Champaign, Urbana IL

Investigators

Abstract

One crucial component that has been missing in automatic control so far is a good understanding of the coupling between control objectives and information constraints. If we view an overall closed-loop control system as a network of information-processing modules, the rate of information transfer between different components of the system in reality is never unlimited, due to constraints on the sensor capability, computational resources, and communication capacity. The increasing interest in high-performance networked embedded systems and autonomous robots puts in great demand a new control theory that can take information constraints into account. The primary goal of the proposed project is the development of a comprehensive performance-based theory of control with limited information. The main contribution of the project will be to address realistic nonlinear models for the control plants and sensors (e.g., those motivated by vision-based control) and to consider control and observer designs that provide guaranteed performance and robustness. Intellectual Merits: The new control theory will reveal fundamental capabilities and limitations of control-theoretic techniques in the constrained information context, thus bringing into better focus the role of systems and control theory in solving today's real-world problems. More specifically, we will verify the new theoretical results in the study of vision-based control of autonomous robots. We believe that the legacy of the proposed project will be a successful marriage of the mathematical control theory with an engineering problem of great practical importance. Broader Impacts: The proposed research will be naturally integrated with education through further improvement of recently developed courses on hybrid systems and control and computer vision, as well as through introduction of new courses on nonlinear control design and identification of hybrid models and systems. The proposers will ensure wide result dissemination and outreach through invited lectures, tutorial workshops, web-based materials, and involvement of students with diverse geographic and ethnic backgrounds in accordance with the policies of the University of Illinois.