« iepriekšējāTurpināt »
University Hospital. (We were told that Gerhardt is also a lecturer in their M.S. A number of the research projects Western research in this area was not program.
were presented at ICCIM'91 and these sufficient for Asian people, who require The list below suggests the range of papers are included in the Proceedgreater flexibility for squatting. Inter- projects that are being studied. ings. A good one (Lim, GINTIC) estingly, the joint model we saw has
describes the control and communicabeen developed in stainless steel and is At the CAD and Knowledge-Based Center tion system designed and implemented very heavy. I would expect that new
at GINTIC beginning in 1985. There is materials research could provide alter- • Parametric feature-based intelligent also a discussion of a C++ graphical native solutions.) Two new industrial component designer and modular toolkit (C. Wah et al.) developed using services have been initiated--printed feature design expert system.
object-oriented techniques. At least one circuit board (PCB) design assembly
paper generated some healthy criticism. and testing and rapid prototyping using • Intelligent knowledge-based Oo L. Tshon described a hierarchical stereolithography. The latter especially process planning system for the approach to mechanical design based was a big hit when demonstrated at the manufacture of progressive dies. upon a tree of tasks, group leaders, and ICCIM91 vendor exhibit and later when
specialists. His approach allowed it was shown to us at GINTIC. Stereo- • Daily production scheduling system communication between specialists lithography (SLA) was developed by for the manufacture of PCBs. within a group but only between leaders the U.S. firm 3-D Systems and uses a
of different groups, and even then only laser on a liquid photocurable polymer • Multilevel neural network job shop those at the same level. Along with to form models directly and very quickly scheduling system.
several other members of the audience, from CAD/CAM data, so called direct
I doubted that this very rigid approach three-dimensional (3-D) manufactur- At the PCB Flexible Manufacturing System would be successful at modeling realing. It is said to be a breakthrough (FMS) and SLA System
world systems. because of the speeds with which pro
We mentioned earlier the AGV totype parts can be produced. SLA • DIP, axial, and radial inserters for demonstration. This is very traditional models are also heavily used for mold through-hole boards.
technology but impressive that it was making, but most of the parts built this
done locally by students. I shared the way are still only good for rapid proto- • Screen printing.
sense of many others that while typing, not real production (due to
researchers in the West should not look insufficient strength). GINTIC has 3-D • System for auto generation of inser- to GINTIC for really new ideas, they Systems' 10-inch unit, allowing direct tion data for PCB assembly line. should be aware that new concepts will forming of parts up to 10 inches on a
be tested here and improvements made, side (the company's largest unit allows • SLA for rapid prototyping. and if deemed useful, will be developed 25-inch parts). Larger parts can be
into working products for local indusbuilt in sections. (Several other U.S. At the Precision Machining FMS try. At the same time a large collection and Japanese groups are also very active
of specialists will be educated in their in this direct 3-D manufacturing • Hierarchical cell, workstation, and technology.)
equipment controller. Education and training are accom
JAPAN-SINGAPORE AI plished by an emerging M.S. program • ASRS workstation controller. (JSAI) CENTER in CIM just begun July 1990. Presently the program has about 40 students. • Robotics modular fixturing work- This facility is housed on one large There are also 2 Ph.D. students, 20 full- station controller.
floor (1,200 m2) of a building within time (M.S.) graduate students, and 68
Singapore's Science Park. It had its part-time graduate students. It should • AGV workstation controller. official opening during the week be mentioned that this program was
ICCIM'91 was in progress, and I took heavily modeled after ones at RPI and Vertical machining center worksta- some time off from the meetings to Fraunhofer and that Prof. Lester tion controller.
visit. The program is viewed as an Gerhardt (RPI), has been active in
important step by both the Japanese collaborating with GINTIC staff in • CMM workstation controller. and Singaporean Governments; the forming the new program. In fact,
Japanese Ambassador to Singapore as
well as the Singaporean Minister for Managers”; a 2-week course, “Intelligent system grinds (for quite a long time) Finance presented speeches at the Systems for Business Professionals”; while various rules are checked for the opening. A short history of the plans and a 14-day course emphasizing more best route based on shortest time and/ and construction of the facility was given detail, “Expert Systems/Knowledge or cheapest fare. I felt that most people via a slide show, and while lighthearted Based Systems for IT Professionals.” could make a similar mental calculathe message was clear; pictures of the In addition, the heart of the program is tion in less time. Some constraints are new center were accompanied by the a 6-month “Prototyping Expert Systems/ allowed, however, and these could make Hallelujah Chorus from Handel's Knowledge-Based Systems” course, such a system really useful, say at an Messiah and the theme music from enabling trainees the opportunity to airport. Unfortunately, the question “2001”!
build prototypes of “real” systems under asked by all Tokyo tourists--how do I Singapore is staking a significant expert guidance. The Singaporean get from here to there without going up fraction of its future on automation; Government is encouraging industry stairs--was not available. Maybe later. the theme is that by the year 2000 participation in a very direct way, by For information contact Singapore will be an “intelligent island.” providing a 70% absentee payroll subsidy To this end the JSAI Center is seen as for company-sponsored employees who Director a way of increasing awareness about AI are either Singaporean or permanent Japan-Singapore AI Center in Singapore, learning needed skills, residents.
CINTECH II developing a friendly base between the This is a "win-win" program as I see 75 Science Park Drive #01-01/04 two countries in a important techno- it for both Singapore and Japan. Others Singapore 0511 logical field, learning about Japanese should view it this way, too, as long as Tel: +65-779-3088 management practices, as well as they appreciate that the emphasis at Fax: +65-779-6162 building relationships and contacts. My the center is on real-world applications-earlier report detailed a number of Al- this is definitely not going to be an The Japan-Singapore Al Center is like projects that have already been academic research facility. Singapore not the only software R&D activity implemented. In total about 50-60 gets specific assistance as outlined above. supported by non-Singaporean comprojects have been developed; one-third Japanese companies get to place their panies as Table 2 shows. The list of are already in use, and the remainder equipment, software, and personnel in applications should be thought of only are in prototype form. There are about Singapore and encourage use of their as representative. In addition, IBM, 200 AI professionals now working in products during training, plus perhaps DEC, Sun, and probably other vendors the country, while about three times get assistance in developing market- have some software development in that number are projected to be needed. able software.
Singapore. Also in Singapore there is a The Al Center will be an important After the formalopening, there was French-Singapore (F-S) Institute and vehicle for accomplishing this goal. an opportunity to browse around the a German-Singapore (G-S) Institute Japan has given the center about Y380M center and discuss various demonstra- on adjoining sites. Although I did not (about $2.5M), mostly in the form of tion projects. These included a Tokyo visit either place, at least one paper on equipment (32 workstations, 25 386-PCs, subway advisor, production line diag- FMS at ICCIM'91 was authored by a 17 Macintoshes, etc.) and software. In nosis system, myopathy consultation scientist from the F-S Institute (Fertin). addition to equipment, Japan provides system, hospital operation theater pri- Prof. Peter Sackett, Cranfield Institute software and technical expertise, while oritization system, traffic data man- of Technology (E-mail: sackett@ Singapore provides facilities, local agement system, production engineer- cim.cranfield.ac.uk), explained to me personnel, and operating expenses. Six ing expert system, pattern optimiza- that the G-S Institute is more closely Japanese Al experts will work at the tion system, etc. Many of them were aligned to manufacturing than the center and five or six Singaporeans running on NECEWS4800 systems. A French, which concentrates on IT issues, will go to Japan for training with few are potentially useful, but others although both illustrate the Singapore various companies (four are already were obviously for teaching. In the lat- approach by providing training, 48-week there). It is also planned that about 500
ter category was the Tokyo subway year, 2,000-h/yr classes in 2- or 3-year training slots will be utilized within advisor. A map of the (admittedly
map of the (admittedly programs leading to a diploma (not a 4 years. In fact, training is a key element complicated) Tokyo system is displayed degree). Some institutions recognize in the center. Three courses are planned: and then users select an origin and outstanding performance on these a 3-day course, “Intelligent Systems for destination station (in Japanese). The
destination station (in Japanese). The programs as equivalent to a good degree
and valid as entry to a masters program. has his Ph.D. from New York Univer- During my visit Nanda showed me Sackett emphasizes that the institutes sity (NYU) (Courant) in functional the computational facilities of his are not research oriented, although they analysis has been retraining himself in department, which are extensive, includare very well equipped and staffed, and applied and computational areas. For ing numerous workstations, a very they do undertake ambitious project several years he has been developing a excellent computerized library card work. One example is the building, again system, running on a PCand written in catalog, and a campus area network. by students, ofa fully operational FMS. APL, for plotting and manipulation of But he was only mildly upbeat about
mathematical entities. His system has a the research activities. He felt that NUS NATIONAL UNIVERSITY number of special functions of mathe- students work diligently, but there are OF SINGAPORE (NUS) matical physics built in, and he is work- very few graduate students and that it
ing on other capabilities such as numer- will take time for the university to While I had visited NUS earlier, I ical integration, solution of differen- develop into a true research instituwas anxious to visit again in order to tial equations, etc. Nanda claims that tion. In short, whatever money can buy see
his targets are students, but in this arena has mostly been purchased; the rest is
there are already a number of good up to the faculty. NUS is encouraging Prof. T.R. Nanda
products, such as Matlab, MathCad, visiting scholars and this will definitely Dept of Mathematics
Mathematica, etc., and so he will have aid the environment. I also had an National University of Singapore a tough time against those, especially opportunity to meet one U.S. visitor 10 Kent Ridge Crescent
as his system is not particularly extensi- spending his sabbatical at NUS, Prof. Singapore 0511
ble. Nevertheless, it is a good piece of Jerome Klotz, from the University of Tel: +65-772-2940
work and he would like to hear from Wisconsin's Statistics Department E-mail: email@example.com potential collaborators and users. At [E-mail: firstname.lastname@example.org). Prof.
the moment he is developing this pretty Klotz echoed Nanda's assessment that Nanda's work on user interface for much in a vacuum and definitely needs NUS's students are hard working, but numerical software was called to my input and suggestions from outsiders. mostly not used to a Western style of attention by Prof. B. Datta. Prof. Nanda
CONCLUDING REMARKS those Chinese students who are gradu. Besides, the Western model is prob
ating from good U.S. or European ably not appropriate in any country in Every visitor to Singapore is universities with doctoral degrees and Asia, and especially those with a impressed with the physical facilities who do not wish to return to mainland Confucian streak that venerates age and equipment. I mentioned this above China. In addition to good financial and family leadership. Singapore is free, in the context of GINTIC and again incentives, they and their family but one doesn't step too far out of line about NUS. But people are worried members are granted immediate here, anyway not if one expects to have about this hardware-heavy/but software- permanent residence in Singapore. a career and a future. weak situation in their country. (Soft- Apparently some initial success of this To repeat, coordination is a strong ware here means particularly human special recruiting program has already point in Singapore, strong leaders + resources.) Some Singaporeans feel that been achieved.
good visions-->well planned actions-they are lacking a pool of experienced, All scientific visitors to Singapore >careful monitoring of action results well-trained people who can take full must also come to grips with the reali- + revision of plans and visions. Sounds advantage of this wealth of hardware ties of the political world here, which is reasonable, and certainly appears to and hopefully lead their country to future not as open as one might find in some
work in this small country. growth. At the lower levels I have other countries. Singapore has been mentioned the emphasis on training run in a very authoritarian way by its ACKNOWLEDGMENT already. At the higher levels this situa- benevolent Prime Minister Lee for many tion is claimed to be particularly bad at years. The system is slowly changing, I wish to note my sincere thanks to the universities, which are struggling and to be fair most Singaporeans and Profs. Lu (University of Illinois), to recruit good faculty members for many outsiders think that restrictions Gerhardt (Rensselaer), Sackett their engineering programs. I was told are reasonable and have benefitted their (Cranfield), and Ho (GINTIC) for that the Government has set up a spe- country. Nanda feels that it is still too reading this report and for their many cial fund and laws to attract to Singapore soon to expect a completely open soci- helpful comments and suggestions.
ety, and perhaps this will never occur.
FIRST KOREA-JAPAN CONFERENCE
ON COMPUTER VISION
The First Korea Japan Conference on Computer Vision, held 10-
by David K. Kahaner
Conference chairmanship was shared describing Japanese and Korean SUMMARY between
research; and (2) the Japanese scien
tists were very gracious to have particThe First Korea Japan Conference Prof. Chung-Nim Lee
ipated so enthusiastically. A preview on Computer Vision was held on Basic Science Research Center of what was going to happen was given 10-11 October 1991 in Seoul's KOEX Pohang Institute of Science and at the opening lectures when research Conference Center. Approximately Technology (POSTECH) in the two countries was reviewed. A 80 papers were presented (31 from P.O. Box 125
literature search on computer vision Korea, 47 from Japan) in three parallel Pohang, Kyung Buk 790-330, Korea uncovered fewer than one-tenth as many sessions during the 2 days. (KOEX, a Tel: +82-562-79-2041
Korean as Japanese research papers. massive convention facility, could easily Fax: +82-562-79-2799
Although the paper balance at the have swallowed 50 such conferences.) E-mail: email@example.com conference was much more even, it was Two invited papers summarized com
very clear that Korean research is at a puter vision research in Korea and Japan, and
much earlier stage and that applicarespectively, and a panel discussed
tions of vision in industry are much “Application of Computer Vision for Prof. Masahiko Yachida
more limited than in Japan. There Automation.” There were about 200 Faculty of Engineering Science were a (very) few exceptions. For attendees. As few Japanese or Koreans Dept of Information and Computer example, Prof. T. Poston in POSTECH's know each other's language, the con
Mathematics Department [E-mail: ference language was English. Korean Osaka University
firstname.lastname@example.org gave an scientists are very far behind in devel- 1-1 Machikaneyama-cho
elegant discussion of the use of singuopment and industrial applications of Toyonaka-shi, Osaka 560, Japan larity theory for vision applications, computer vision. In basic research they Tel: +81-6-844-1151 x4846
but at the moment this work is very far are lagging a bit less because many Fax: +81-6-845-1014
from practical realization. Also, some Korean scientists have done research E-mail: email@example.com Korean industry is using vision techwork abroad in the United States and
niques and, in particular, Samsung has Japan and are attempting to continue I had met Prof. Lee almost 20 years a visual soldering inspection system that in Korea. Both countries see anal- earlier when we were both in the Math- that has many similarities to systems ysis of moving images as the most ematics Department at Ann Arbor. He developed by Toshiba and NTT. important new research area, although was trained as a pure mathematician Computer vision is usually thought many “standard” topics are far from but is now interested in applied prob- of as beginning with a real world situasolved. (The recent interest in move- lems, particularly those concerned with tion (scene) that is input through a ment stems from increasing processing computer vision.
camera and then digitized and ending speed. Without fast computers and fast The fairest things that can be said with a description of the scene, e.g., data movement, researchers had to about this conference are that (1) the knowledge (perhaps leading to action sample images every second or two, Korean scientists were very brave to which changes the scene, etc.). The inleading to very complicated relations have organized it, especially to have between steps are often modularized. between successive frames.)
scheduled more or less alternating talks Typically there is a "camera model”