Skip to article sections

Intelligent Control Systems: An Introduction with Examples

Katalin M. Hangos, Rozália Lakner and Miklós GerzsonKluwer Academic PublishersDordrecht2001xvi + 301 pp.ISBN 1-4020-0134-7hardback, $92.00, £63.00, 100.00Applied optimization series, No. 60

This is intended as a teaching text and its contents have been the basis of a course in the University of Veszprém, Hungary, over 5 years, taken by more than 100 students. It is also suitable for self-study and is claimed to be the first comprehensive textbook in this interdisciplinary field.

The treatment begins with an essentially AI flavour, with discussion of knowledge representation in databases and rule sets, and of the associated reasoning and search. Attention is given to verification and validation of rule-bases, where the properties to be checked are free from contradiction, and completeness, the latter in the sense of always coming up with an answer.

The AI languages LISP and PROLOG are treated quite thoroughly, as well as expert system shells, and attention is given to the special requirements of expert systems to operate in real-time, where lock-out must be avoided. Apart from total lock-out, harmful processing delays are avoided by assigning levels of priority to the respective queries, and allowing an urgent query to interrupt lower-priority processing.

The final four chapters treat specific topics. One is on Qualitative Reasoning and introduces imprecise methods depending on sign and interval calculi. These methods appear to correspond to human “rough and ready”ways of thinking about control problems, methods that are effective in deciding roughly what will happen a short time ahead or in tracing out a tree of possible futures.

Two chapters treat, respectively, Petri Nets and Fuzzy Control Systems, and both give very clear introductions. The final chapter is on the implementation of a real-time expert system using a commercial shell called G2, from Gensym Corporation. This is extremely comprehensive and allows interaction through a graphical representation of the process that is produced by the shell. It includes the means of running a process simulation, either to check the validity of a policy before going online, or to run high-speed simulations to help determine control actions.

A defect of the book, probably remedied in the taught courses, is that in two respects it fails to fill in a wider picture. It is not made clear just how the topics fit together, and for instance how those of Qualitative Reasoning and Petri Nets fit in. It is not clear whether it is intended that these should be incorporated in real-time control systems (perhaps by specially bright students at some future date) or whether they are included as aids to manual analysis. The wider picture of expected application area is also not filled in. An indication of what the 100+ graduates went on to do would be interesting to readers as well as to prospective students in Veszprém. The book’s title promises “with examples” but the only practical example treated is a simple one obviously devised for expository purposes.

The treatment is firmly focused on practical applications, but there is much here that could interest psychologists and neurophysiologists, at least in suggesting lines of enquiry. Most of the activity of living creatures has greater correspondence to the operation of a real-time controller than it has to the mathematical and logical problem solving that has been the focus of much AI effort. The apparently seamless merging of continuous control with strategic decision making, as displayed for example in the playing of ball games, is still a special feature of living systems. The considerations of priority in querying a rule-based system, discussed in the book, bring to mind the “administrative methods” outline in early reviews of AI by Marvin Minsky (1959, 1963) and this could be just one way in which attention to real-time control can draw attention to basic issues that have tended to be ignored.

Attention to these very general issues could occupy another volume, and the book as it stands is a welcome addition to the literature that will be much in demand, both for study in its entirety and for its relatively painless introductions to the separate topics.

Alex M. Andrew

Minsky, M.L. (1959), “Some methods of artificial intelligence and heuristic programming”, Mechanisation Thought Processes, HMSO, London, pp. 3-36.
Minsky, M. (1963), “Steps toward artificial intelligence”, in Feigenbaum, E.A and Feldman, J. (Eds), Computers and Thought, McGraw-Hill, New York, pp. 406-50.

Data & Figures

Supplements

References

Minsky, M.L. (1959), “Some methods of artificial intelligence and heuristic programming”, Mechanisation Thought Processes, HMSO, London, pp. 3-36.
Minsky, M. (1963), “Steps toward artificial intelligence”, in Feigenbaum, E.A and Feldman, J. (Eds), Computers and Thought, McGraw-Hill, New York, pp. 406-50.

Languages

or Create an Account

Close Modal
Close Modal