Intelligent Control: An Overview of Techniques

Abstract

In many established fields, the label ?>intelligent ?> heralds new developments that take issue with some traditional assumptions in research. In the case of intelligent control, an explicit attempt is made to draw inspiration from nature, biology, and artificial intelligence, and a methodology is promoted that is more accepting of heuristics and approximations???-???and is less insistent on theoretical rigor and completeness???-???than is the case with most research in control science. Beyond such general and abstract features, succinct characterizations of intelligent control are difficult. Extensional treatments are an easier matter. Fuzzy logic, neural networks, genetic algorithms, and expert systems constitute the main areas of the field, with applications to nonlinear identification, nonlinear control design, controller tuning, system optimization, and encapsulation of human operator expertise. Intelligent control is thus no narrow specialization; it furnishes a diverse body of techniques that potentially addresses most of the technical challenges in control systems. It is also important to emphasize that intelligent control is by no means methodologically opposed to theory and analysis. Chapter 6 of this book, for example, discusses some theoretical results for neural networks and fuzzy models as nonlinear approximators Introductory tutorials to the key topics in intelligent control are provided in this chapter. No prior background in these topics is assumed. Examples from ship maneuvering, robotics, and automotive diagnostics help motivate the discussion. (Other chapters in this volume, notably Chapter 16, also outline applications of intelligent control.) General observations on autonomy and adaptation???-???two characteristics that are often considered essential to any definition of intelligence???-???are also included.