Fifty Years of Information Based Control Theory

Abstract

The year 1948 was auspicious for information science and technology. Norbert Wiener's book Cybernetics was published by Wiley, the transistor was invented (and given its name), and Shannon's seminal paper "A Mathematical Theory of Communication" was published in the Bell System Technical Journal. In the years that followed, important ideas of Shannon, Wiener, Von Neumann, Turing and many others changed the way people thought about the basic concepts of control systems. Hendrik Bode himself was a Shannon collaborator in a paper on smoothing and prediction published in the Proceedings of the IRE in 1950. It is thus not surprising that by the time the earliest direct predecessor of CDC (the Discrete Adaptive Processes Symposium) was held in New York in June, 1962, concepts from machine intelligence and information theory were not at all foreign to the control community. 

This talk will examine the interwoven evolution of control and information over the past fifty years during which time the IEEE Conference on Decision and Control went from infancy to maturity. The talk will also discuss two new areas in information based control. In collaboration with W.S. Wong, some recent work on control communication complexity has been aimed at a new class of optimal control problems in which distributed agents communicate through the dynamics of a control system in such a way that the control cost is minimized over many messages. Applications of the theory to robot communication through relative motions (e.g. robot dancing and team sports) and to distributed control of semi-classical models of quantum systems will be discussed. The talk will also discuss some recently discovered links between information and the differential topology of smooth random fields (joint work with D. Baronov). The latter work has been applied to a problem of rapid information acquisition in robotic reconnaissance, and it has suggested metrics by which to assess the tradeoff between speed and accuracy.