The data-driven approach to classical control theory

A data-driven approach to control design has been developing, since the early 1990's, upon the concepts and the methods of classical control theory; to this approach we refer as data-driven classical control. This is now a consolidated theory, with a large body of methods and a great number of applications. In this paper we present a survey of these developments along the past three decades. The theory is overviewed, various methods are described, their applications are illustrated and some issues that are sill open for future research are discussed. The extension of these concepts to the design of nonlinear controllers is an emerging and promising subject that is also discussed, and it is shown how it can be approached by the theory of geometric control and the tools of machine learning.