Site: Tokyo Institute of Technology
Department of Systems Science
4259 Nagatsuda, Midori-ku
Yokohama 226, Japan
Date Visited: May 25, 1995
Report Author: J. Foley
Professor Michio Sugeno
Professor Kaoru Hirota
Dr. Anca Ralescu
Professor Michio Sugeno's lab at the Tokyo Institute of Technology has been conducting research in fuzzy logic for many years. In addition, he was instrumental in forming LIFE (Laboratory for International Fuzzy Engineering Research), a MITI-sponsored and funded fuzzy logic research collaboration that involved roughly 50 companies. The LIFE project recently ended, as planned, after existing for six years. Dr. Anca Ralescu was associate director of LIFE and is spending several months with Dr. Sugeno before returning to her home institution, the University of Cincinnati. Fuzzy logic is of relevance to human-computer interaction because it represents a means to interface between the terminology of everyday discourse (a little warmer, a lot colder, a bit more to the left) and the preciseness of computers (temperature = 1.01 * temperature; heading = heading - 5). The author's hosts knew of no commercial user-computer interfaces in which fuzzy logic is used. A number of prototype interfaces were developed as part of the LIFE project, and OMRON has a prototype fuzzy extension to SQL for database queries (see OMRON site visit report for more details). While fuzzy logic was initially developed in the United States by L. Zadeh of the University of California at Berkeley, it has been further developed and used much more extensively in Japan. When asked why fuzzy logic has been more popular in Japan than in the United States, Professor Sugeno suggested that Asian cultures are more comfortable with uncertainty and ambiguity than are Western cultures.
Fuzzy logic is taught in about ten undergraduate programs and in many more graduate programs at universities in Japan, mostly in mechanical engineering and electrical engineering departments.
Professor Sugeno described the following applications of fuzzy logic:
Professor Sugeno assisted Yamaichi Securities, the fourth largest brokerage firm in Japan, in building an 800-rule fuzzy logic expert system for stock analysis. One part of the system is used to derive a set of fuzzy logic rules that will predict the future behavior of an individual stock. Input is the stock's performance over a several-year period: output is a minimal set of fuzzy logic rules that explain the stock's behavior over that time period. Given that the stock's fundamentals do not change, the rules also predict future price moves. The set of rules is minimal in that it is the smallest set of rules that explain the stock price moves to within a definable error bound. The system is in use: Professor Sugeno reported that it has performed reasonably well, but did not predict a recent drop in stock prices, as that was caused by a change in fundamentals that were not accounted for in the fuzzy rules.
The Mitsubishi Gallant has an automatic transmission with one position for forward, in contrast to the normal two or three. The fuzzy logic controller downshifts automatically for braking, including on steep hills. The Gallant uses neural net technology to compute the membership functions for the variables used in the fuzzy rules, thereby adapting to the driver's behavior. The Honda Accord also uses a fuzzy logic controller for its automatic transmission, while retaining multiple positions for forward.
Up to 20% energy savings have been demonstrated with fuzzy logic controllers for air conditioners. They are used in several commercial air conditioners, in lieu of the more traditional on/off controller.
The new subway in the city of Sendai, operational since 1987, uses a fuzzy logic running and braking system (Hirota 1987). One new line of the Tokyo subway system also uses such a system.
Professor Sugeno showed a videotape to illustrate one of his projects, a 3.6-meter autonomous helicopter that has a fuzzy logic controller. The helicopter can respond to voice commands (turn left, stop, back up, slow down, speed up, land, etc.) or can fly a predefined course (using a differential global positioning system to know where it is). The helicopter appeared to be quite stable in conducting its maneuvers. Dr. Sugeno is working with United Technologies on an unmanned helicopter.
Hirota, K., ed. 1993. Industrial applications of fuzzy technology, 263-69. Tokyo:Springer-Verlag.
Ralescu, A. L., ed. 1994. Applied research in fuzzy technology. The Netherlands: Kluwer Academic Publishers, ISBN 0-7923-9496-8.