Automatica's
Editor George Axelby asked me to write up my perspective on the interchange. What follows, then, is an edited transcript of my notes rearranged to group topics more naturally, and with my later comments enclosed in [...]. We hope it will stimulate healthy discussion about trends in automatic control.
BRIDGING THE GAP Friedland lamented "the gap" between theory and practice---decrying the fact that eng/neers in industry find the IEEE Transactions on Automatic Control almost impossible to read---and expressed his doubt that even the Editor could read more than one paper per issue. Replying, the Editor, S. Kahne, stressed the high refereeing standards of the Transactions, and the fact that it seemed unwise to suppress good theoretical papers simply because their applicability was not immediately apparent. [One might add that the control community now supports several excellent journals, and it helps readers to know that each has its own distinctive mix of articles.]
Friedland added that the "soldiers" whose formal education included Bode plots and other techniques of twenty years ago cannot use modern control techniques, and urged modern theorists to get out and teach how to tackle real world problems. [The chalienge to the universities is to make abstract concepts accessible to the most 'hard-boiled', engineer, while avoiding the temptation to present these concepts so 'elegantly' that few students have a real feel for their origin or applicability. We probably need courses which integrate theory with detailed cases studies---including a guide as to when it's better to use an off-the-shelf component than to design one from scratch. Should industrial internships become a required component of graduate education in control?] Kokotovic felt that there was not so much a gap between "classical" and "modern" as there was a need to balance the range of abstractions. The basic problem of much theory is that young workers add beauty and generality without the experience of working with specific systems. Unfortunately, many real world problems are so unstructured that a description of them would yield papers as unreadable as modern theory papers. Somehow, one has to combine practical experience with the tools afforded by theory--he recalled Rosenbrock's observation that the control of a 75th-order system is almost intractable unless one has such structural information as that it is a distillation column. Nichols observed that it is the availability of sensors (measuring what is to be controlled) and actuators (which can mediate the control) which structure the abstract equations.