Modeling and simulation of robot manipulators: A Parallel processing approach: by Albert Y. Zomaya; World Scientific Series in Robotics and Automated Systems; World Scientific; Singapore; 1992; xvi + 295 pp.; $48; ISBN: 981-02-1043-4
✍ Scribed by Janusz Zalewski
- Publisher
- Elsevier Science
- Year
- 1995
- Tongue
- English
- Weight
- 437 KB
- Volume
- 3
- Category
- Article
- ISSN
- 0967-0661
No coin nor oath required. For personal study only.
✦ Synopsis
The titles of the respective sections speak for themselves---knowledge-based design, the fuzzy controller as a variable structure system, and sliding-mode control all look at the analysis of the controller from the point of view of control engineering.
Numerous defuzzification schemes are covered next, where the authors discuss several possible ways of converting a fuzzy set into its numerical "equivalent". The material includes research outcomes obtained by the authors themselves (BADD transformation, M-SLIDE, the RAGE approach, and others). Finally, the end of the book is devoted to flexible structures in fuzzy modelling. In brief, the main idea is to develop the calculus of fuzzy rules used in system modelling in the setting of the generalized aggregative operations. This defmitely adds a lot of flexibility to the standard Mamdani-type of fuzzy-rule calculus.
In terms of its content, the book is carefully arranged, lucidly written, and full of well-selected identification algorithms, analyzed in depth. There are, however, two essential modelling and control issues that have been left out. The first concerns fuzzy model validation and verification. Is the model valid?
Should it be rejected? What are the confidence intervals associated with the output of the model? These questions, well posed in the setting of regression analysis supported by statistical tools, should be reformulated in the language of fuzzy sets. While some ideas have been proposed and studied earlier, there is still a long way to go. This book does not tackle this aspect at all. The second point concerns the links between fuzzy models and control algorithms. Interestingly, as these two constructs are rule-based, the design of the controller could be conveniently carried out at the linguistic level, not even getting down to the numerical stratum via the defuzzification method. In this sense, all the design activities concentrate on the development of a suitable rule-based structure (controller) capable of cooperating with the given one (fuzzy model).
Another point to be discussed deals with the defuzzification step---an element that is unique to fuzzy modelling and not encountered in any other algorithms of standard numerical model-building. While the problem statement is correct from a formal point of view, the semantics of the defuzzification problem are not that obvious and fully convincing. The problem, as it has been set up (not only in this book but in the current literature) becomes to some extent 'Tart pour l'art". The main weakness stems from the fact that the defuzzification machinery cannot be divorced from the overall methodology of fuzzy identification and the construction of the fuzzy model itself. The above, somewhat general and critically constructive comments do not refer only to this book, but allude to the area of fuzzy modelling in general.
What is much more important, though, is that the authors convey a clear message: fuzzy modelling is a mature discipline with its own methodology and design tools. In summary, the book is an outstanding and fully updated source of valuable algorithms of fuzzy modelling, and is strongly recommended to everybody interested in investigating new frontiers in system modelling, and of significant value to those pursuing new frontiers of intelligent control.