Site: Omron Corporation
Kyoto R&D Laboratory
Shimokaiinji, Nagaokakyo-City
Kyoto, Japan
URL: http://www.omron.co.jp/home-e.html

Date Visited: May 23, 1995

Report Author: J. Hollan

ATTENDEES:

JTEC:

S. Chipman
J. Foley
E. Glinert
J. Hollan
R. Kraut
T. Sheridan
T. Skelly

Hosts:

Masoat Kawade
Tel.: 81-75-953-3880
Fax: 81-75-952-0411
E-mail: kwade@zoo.ncl.omron.co.jp
Taiji Sogo
Tel.: 81-75-953-3880
Fax: 81-75-952-0411
E-mail: sgh@zoo.ncl.omron.co.jp

BACKGROUND

Omron was established in 1933. At the time of the JTEC visit, the Omron Group employed 22,190 people and had net sales approach 400 billion (~$4 billion). The corporate philosophy is: "To the machine the work of the machine; to man, the thrill of further creation." The company focuses on computers, communications, and control. Control components and systems account for about 60% of net sales.

The computer program started in 1981. In addition to large numbers of communications and control products (for example, more than 100,000 factory automation components), Omron also produces a 32-bit 100 mip Unix-based workstation (jointly with Data General) and a digital fuzzy processor (FP300; it operates at 10 million fuzzy logic processes per second).

RESEARCH AND DEVELOPMENT ACTIVITIES

Research and development activities are focused in four areas: microcomponents, computers-communications-control, life sciences, and fuzzy logic. About 8% of revenue is spent on R&D. Research in fuzzy logic technologies has been ongoing since 1983; about 25% of products incorporate these technologies. Fuzzy-related products produced 55 billion (~$550 million) in 1994 revenue, and sales are expected to reach 100 billion (~$1 billion) in 1995. The demonstrations that the JTEC team saw all involved applications of fuzzy logic. The Omron researchers contrast fuzzy logic methods using membership functions with what they term crisp logic and linear-methods. There are approximately 100 researchers in the company's central lab working on fuzzy logic and applications. Omron has about 1,000 patents in the area of fuzzy logic. Company researchers are working in five main areas: knowledge base control (e.g., temperature controller); sensing with recognition (e.g., vibration security sensor); image recognition (e.g., white blood cell recognition); knowledge information processing (e.g., fuzzy database); and automobile applications.

The first demonstration was of an image understanding system. It was running on a Sun with a frame-grabber and Verbex Speech Commander speech recognition system. The system identifies objects in a typical office setting and answers by voice. The system uses fuzzy logic and Dempster-Shafer in the image identification system. It responds to spoken questions. For example, when the operator says "Where is the telephone?" the system responds with "Something like a telephone is around left on the desk." They see potential applications in the areas of security systems, remote operation of industrial robots, disaster prevention systems, and navigational support for the blind. The image recognition system makes inferences based on fuzzy representations of color, size, aspect ratio, and location.

The second demonstration was of an application of fuzzy logic in database applications. Omron has developed a Fuzzy extension of the SQL data base query language. Users can give a query like "young and tall person with annual income more than about $50,000." Fuzzy-SQL operates on top of a traditional database. Direct Mail is one potential application that company representatives mentioned.

SUMMARY

Work at Omron is focused on applications in their main product area: control components and systems. Omron was the first to introduce ATM machines in Japan, and is involved in many other areas, such as traffic control systems, point of sale terminals, automated ticket vending machines, factory and manufacturing automation, communication components, and health care systems. A large number of components involve application of fuzzy logic. The company also produces a high-speed fuzzy logic controller, a compact fuzzy logic controller, a fuzzy logic module for PCs, and a general purpose fuzzy-logic-based workstation.


Published: March 1996; WTEC Hyper-Librarian