Oral-History:Yoshiaki Shirai

About Yoshiaki Shirai

Yoshiaki Shirai, a Japanese mechanical engineer and university professor, was born on 3 August 1941 in Toyota, Japan and is the son of Takeaki and Yoshiko (Sasaki) Shirai. He earned a B.Eng. degree in Mechanical Engineering from Nagoya University in 1964, and the M.Eng. and the Dr.Eng. degrees in Mechanical Engineering from Tokyo University in 1966 and 1969 respectively. Shirai is an inventor, innovator, and leader in the fields of robotics artificial intelligence, computer intelligence, and system control. His research projects have included hand posture recognition, image processing, and sensors for surgical training.

Shirai has been an active member of IEEE; the Robotics Society Japan; the Japanese Society of Artificial Intelligence; Information Processing Society of Japan; the Institute of Electronics, Information, and Communication Engineers; the Robotics and Automation Society, and the American Association of Artificial Intelligence. In addition, he served on the board of directors of the Japan Society Artificial Intelligence from 1985 to 1989 and the Information Processing Society Japan from 1987 to 1989.

In this interview, Shirai discusses his education and training as a mechanical engineer and his career in robotics and artificial intelligence.

About the Interview

YOSHIAKI SHIRAI: An Interview Conducted by Selma Šabanović with Matthew Francisco for the IEEE History Center, 30 August 2011.

Interview #800 for Indiana University and the IEEE History Center, the Institute of Electrical and Electronics Engineers Inc.

Copyright Statement

This manuscript is being made available for research purposes only. All literary rights in the manuscript, including the right to publish, are reserved to Indiana University and to the IEEE History Center. No part of the manuscript may be quoted for publication without the written permission of the Director of IEEE History Center.

Request for permission to quote for publication should be addressed to the IEEE History Center Oral History Program, IEEE History Center at Stevens Institute of Technology, Castle Point on Hudson, Hoboken, NJ 07030 USA or ieee-history@ieee.org. It should include identification of the specific passages to be quoted, anticipated use of the passages, and identification of the user. Inquiries concerning the original video recording should be sent to Professor Selma Sabanovic, selmas@indiana.edu.

It is recommended that this oral history be cited as follows:

Yoshiaki Shirai, an oral history conducted in 2011 by Selma Šabanović with Matthew Francisco, Indiana University, Bloomington Indiana, for Indiana University and the IEEE History Center, Hoboken, NJ, USA

Interview

Interviewee: Yoshiaki Shirai

Interviewer: Selma Šabanović

Date: 30 August 2011

Place: Flagstaff, AZ Time: 1:42:42

Education and Training

Q:

If we could start with your name, where, and when you were born?

Yoshiaki Shirai:

My name is Yoshiaki Shirai. I was born in 1941 in Toyota, Japan.

Q:

Where did you go to school?

Yoshiaki Shirai:

Do you mean, school, university, or-

Q:

Even when you were younger.

Yoshiaki Shirai:

My high school was Tokai High School in Nagoya. I also attended Nagoya University because I was living in Nagoya. I transferred to Tokyo University for a graduate course and spent five years there, finishing with both a master’s degree and a Ph.D.

Q:

What did you study? What department were you in at Nagoya University?

Yoshiaki Shirai:

Mechanical engineering.

Q:

How did you get interested in mechanical engineering?

Yoshiaki Shirai:

I was born in Toyota. There is a Toyota factory, so I was interested in cars. Many people in that area would like to go into mechanical engineering. [Laughter.]

Q:

It was just popular. [Laughter.]

Yoshiaki Shirai:

Yes, it was popular. I guess so. I think it was very popular.

Q:

Did your family work for Toyota?

Yoshiaki Shirai:

Yes, my father worked for Toyota, and then he moved to another company, Denso, adjacent to Toyota.

Q:

That’s when you went to Nagoya?

Yoshiaki Shirai:

Yes, it is.

Q:

When you were an undergraduate, did you know about robots?

Yoshiaki Shirai:

Only by comic books.

Q:

Which comic books?

Yoshiaki Shirai:

Atom.

Q:

Did you enjoy reading it?

Yoshiaki Shirai:

Yes, very much.

Q:

How did you decide which university to go to for your Ph.D.? Yoshiaki Shirai:

I stayed in Nagoya for many years for high school, music school, private school, and my undergraduate course, so I wanted a change of place. I think at that time, Tokyo University was known to be the best place for it [mechanical engineering].

Q:

You were in mechanical engineering at Tokyo University.

Yoshiaki Shirai:

Yes.

Q:

Did you go into a lab when you went to Tokyo University?

Yoshiaki Shirai:

In Tokyo University – in every university- we have lectures. We also belonged to a laboratory. When I was studying for my master’s degree, I was in Masahiro Mori’s laboratory. He is very famous for robotics. When I entered his laboratory, he demonstrated his first robot. He said, “this is the first robot in the world that has three fingers and manipulates objects by pneumatic actuators.”

Q:

What research was Professor Mori interested in at the time?

Yoshiaki Shirai:

My field is control, and Mori was interested in control for manufacturing plants. He was especially excited about the automatic production line for the cigar maker near the laboratory, so he stayed outside rather than stay in.

Q:

Oh, so you worked with him on that project?

Yoshiaki Shirai:

No, he said, “Please choose; find your own thing.” I didn’t need him often, maybe several times a year.

Graduate School and Research in Sensors, Robotic Fingers, and Vision

Q:

How did you choose your theme?

Yoshiaki Shirai:

Mori showed me the three-finger robot, so, at first, I decided to continue to make better fingers. At that time, the very famous professor, Ichiro Kato, was an associate professor at Waseda University. I looked at the newspaper, and it said, “Kato Ichiro makes a tactile sensor.” So, I called Ichiro and said, “I’m a student, but could you show me that tactile sensor?” I later visited his laboratory, and he showed me the tactile sensor. It consisted of a small micro switch. After looking at it, I thought it was not a good tactile sensor. Then, I tried to make better sensors.

Q:

What kind of different sensors did you develop? What did you try?

Yoshiaki Shirai:

I tried rubber and carbon powder. I made it, but it was not successful because its characteristics of conductivity change with time. I gave up, and I did many things on fingers. At last, I had more interest in vision. Then, I started to work on vision because we had sensors. We have a very small degree of freedom compared to the retina. It’s a very large degree of freedom in vision. I think that is more promising.

Q:

At that time, was anybody else working on vision?

Yoshiaki Shirai:

No one in my field.

Q:

You had to find everything yourself.

Yoshiaki Shirai:

Yes.

Q:

How did you do that?

Yoshiaki Shirai:

I read books.

Q:

Was it easy to find books at the time?

Yoshiaki Shirai:

No, I had to go to the library.

Q:

What were the books? Who were the authors? Were they Japanese authors?

Yoshiaki Shirai:

The authors were mostly American. <inaudible 00:07:10> I went to the library-the biggest library, the National Diet Library [Japan’s national library in Tokyo]. Sometimes, I would visit there to find the books. There is also an organization, which introduced the foreign journals very quickly. That gave us a summary of the journals. Together with the original, it was okay.

When I was a student, I earned some money by reading the foreign journals and writing the summary in Japanese. Since I registered that my field was vision, I could get the information very quickly and get some money by writing the summary in Japanese.

Q:

That’s very clever.

Yoshiaki Shirai:

If I visited that organization, of course, I could have read more summaries written by other researchers.

Q:

While you were enrolled in your master’s program, what kinds of questions were you working on in vision?

Yoshiaki Shirai:

I was interested in how it can recognize objects.

Q:

What were some of the early studies or early experiments in vision?

Yoshiaki Shirai:

At the time, there were handwritten numbering. Numerical character reading was just established by the Toshiba company for postcard [postcode] classification. We have a postcard [postcode] with numbers in Japan. However, maybe in the United States, it is automatically read and then classified. It was already complete. In the United States, there is also a system, which analyzes traces of the high-speed particles, half-automatic. It’s interactive.

Later, when I visited in the United States and went to a laboratory, I looked up that machine. It was mostly automatic, but if they made a mistake, they had to go there, point it [the mistake] out, and find it [the solution]. This kind of machine was already completed.

Q:

What kinds of experiment did you do during your master’s program?

Yoshiaki Shirai:

During my master’s program, finding the theme took one year. Then, in not so much time, I found a very interesting book called Perceptron. 1964. I found that is interesting, and I tried to apply or improve the performance of the perceptron [algorithm for binary classifiers].

Q:

Were you still working by yourself?

Yoshiaki Shirai:

No.

Q:

No. Whom did you work with then?

Yoshiaki Shirai:

After I worked for one year, I found that perceptron has a limitation. While reading many other books, I found that this approach was not very good. It changed my direction.

Q:

What did you work on? How did you decide to work on vision?

Yoshiaki Shirai:

I enrolled in a Ph.D. program. At that time, I was working in quite a different field called operation research, but it was while since I was working for my Ph.D. The operation research was for forestry operations, so I visited the laboratory for forestry. There they were <inaudible 00:11:35> paper that is in the analyzer – aerial photograph analyzer was installed inside of the laboratory in that.

I visited that laboratory and asked a related person to introduce me and to let me use that one [analyzer]. I had a dream to analyze this X-ray picture. It’s a mechanical engineering course. I went to the ship building company. They said that they took many pictures for the welding part to see if there were some defects. Human beings analyze this, but I thought that should be analyzed by a machine. I asked if that’s okay.

I left that X-ray picture by that machine, and then some curves came. I then went to the seismic laboratory for earthquake studies in Tokyo University to analyze this curve in the tape – a digital tape. I brought this to the computer center in my university, and that converted the tape to a card. The dividing card is the main input. That resulting card that I worked on recognized defects from an X-ray picture alone. It’s quite different from Ph.D. work. When I finished the Ph.D. there and before I contributed to the Pattern Recognition, Volume I was published.

Q:

Congratulations. [Laughter.]

Yoshiaki Shirai:

Yes. My Ph.D. work is not related to vision. I entered the Electrotechnical Laboratory [ETL] as [Hirochika] Inoue and started working with robotics again.

[Electrotechnical Laboratory (ETL), Ministry of International Trade and Industry (MITI), 1-1-4 Umezono, Tsukuba-shi, Ibaraki 305, Japan]

Electrotechnical Laboratory (ETL) and Robot Vision Group

Q:

When did you enter the Electrotechnical Laboratory?

Yoshiaki Shirai:

I entered the laboratory just after finishing my Ph.D. at Tokyo University.

Q:

So, 1969?

Yoshiaki Shirai:

Yes, it was 1969.

Q:

What kind of work were they doing in ETL?

Yoshiaki Shirai:

There were two related fields. One was robot, mainly interested in manipulation. The other was related to vision. In that laboratory, the majority of work was robot and cat vision.

Q:

Why use a cat?

Yoshiaki Shirai:

The cat is used for the analysis of visual processing.

Do you know [David Hunter] Hubel and [Torsten Nils] Wiesel in Massachusetts? I met them, and they received the Nobel Prize in the analysis of cat retina and the visual activity of cats. In many places, not only Harvard, cats are a typical subject for analysis because the shape of their heads are almost fixed. It’s easy to analyze. I belonged to that laboratory and the robot vision group.

[Hubel and Wiesel received the 1981 Nobel Prize in Physiology or Medicine “for their discoveries concerning information processing in the visual system;” the prize was shared with Roger W. Sperry.]

Q:

You were working on both the cat and the robot?

Yoshiaki Shirai:

Yes.

In the laboratory, I belonged to the robot vision group.

Q:

What was the aim of the robot vision group? What were they doing?

Yoshiaki Shirai:

Maybe it was the same in the United States. When I entered Tokyo University, my professor showed me the movie of the Salford robot. He picked up this block and put it like this with a taped TV camera.

Q:

The blocks world was a record.

Yoshiaki Shirai:

Yes. When we were students, we knew there was such a kind of an intelligent robot. Of course, industrial robots were important in Japan around 1964. They were important already. My boss in the Electrotechnical Laboratory, who was in my section, visited MIT in the United States, and he brought back some pictures of MIT’s work, including block world. We knew what approach to take. We knew what to do. <inaudible 00:17:27> It is similar, I think.

Q:

You were also doing block and kind of pick and place?

Yoshiaki Shirai:

Yes. The vision for pick and place. Pick and place was another section.

Q:

Oh, the manipulation section.

Yoshiaki Shirai:

[Hirochika] Inoue was in that section.

Q:

What were some of the questions or problems that you worked on in the lab?

Yoshiaki Shirai:

At first, we worked on how to make the line drawing of the block world. Then, I found that it was impossible because it depended on the lighting condition. In principle, there’s no difference of light intensity. It was then impossible to find this <inaudible 00:18:19>. Afterwards, I thought of a way to make better drawings that use the four different types of light: first, you use front light, then upper light, then left/right and right/left.

Then we make the tap light each line drawings. In some line drawings, some part is missing and some part is shadow cast because there are many -- <inaudible 00:18:52 > with blocks.

This way, some part is unknown, and another is shadow casted. Combining four line drawings, we can make the perfect one.

At first, I worked for that [line drawings on the types of light]. I thought that was not enough, because at that time, line drawing was okay. After we get the line drawings, you would ask, “How do you interpret as a 3D object?” That presents other issues. I graduated with experience in mechanical engineering, and I thought about how to measure the distance. Then, I thought of a better way because we were using an intelligent camera. We just project straps, scan, and in 3D data, we can get the entire field of view. <inaudible 00:18:52 >.

My purpose was not to make that device. My purpose was to use the three pieces of data to interpret the scene. Then, we just succeeded. At that time, my boss’s boss told me that this was a good chance to go abroad and spend one year in some laboratory. Because my friend was at MIT, I decided to go to MIT. I wrote a letter to Marvin L. Minsky with the two papers for light, the illuminations paper, and the rangefinders paper. They were submitted to each guy in London in 1971. Minsky then accepted me. Later, I knew that Minsky received many letters from the world. It’s not easy to be accepted, and I was very lucky. Then, I just made a good rangefinder, and we could take data. It was now my time to use that data. It was very good timing, but I had to leave this; I had to go to MIT.

Q:

Who was your boss at ETL [Electrotechnical Laboratory]?

Yoshiaki Shirai:

My direct boss was Saburo Tsuji. Later, he became a professor at Osaka University and the president of the faculty. Last year, he received an award from the government. He was also the president of the Artificial Intelligence Society in Japan as well as the vice president of the Robotics Society of Japan. That history is just the same as it was for me.

Q:

You had to leave your work with the rangefinder. Other people were working on it, and you went to MIT.

Yoshiaki Shirai:

Yes.

One Year at MIT

Q:

When you got to MIT, what was happening there?

Yoshiaki Shirai:

Minsky showed me the laboratory. I was told this was my room, and after that, we had no relation. I had to find the theme. One day, I found a camera, <inaudible 00:22:59> an image dissector camera, and asked the students, “What is this?” He, a student or staff, showed me how to use it, and he showed me the performance of the Binford-Horn LINE-FINDER.

[Berthhold K.P. Horn and Thomas O. Binford] are very famous in the field of vision. I found that the performance was not so good. It was poor. Looking at it, I figured I could make it better with line drawings. That was the start of my work at MIT.

Q:

Did you work by yourself or with somebody? Were there any collaborators?

Yoshiaki Shirai:

No, I was alone.

Q:

Inoue-Sensei was there too at the time.

Yoshiaki Shirai:

No, at that time, he was a student at Tokyo University.

Q:

Ah, okay. So, he had been there and went back, and then you went – Or, how did it work?

Yoshiaki Shirai:

Sorry. No, he entered the U.S. one year later. Then, I moved to MIT. He was not a student; he was at ATTI [Advanced Tracking Technologies Inc.]. He’s in another section.

Yes, my idea was that the line drawing was to connect the point where the light intensity changes. If we took a picture of the blocks, some parts where light intensity changes were not so large, and in some parts, there was a small shadow light intensity <inaudible 00:24:39> large. If you point that intensity and then you just connect, you cannot get a better line drawing. Therefore, when we look at something, some part is very clear and something is not clear. At first, human beings find the clear part, and using that as a clue, we find the obscure parts as well as details. So with that idea, I used the findings from the line drawings.

Q:

This was at MIT with the new image dissector camera.

Yoshiaki Shirai:

Yes.

Q:

Did you publish that anywhere?

Yoshiaki Shirai:

Yes, I published in the Journal of Artificial Intelligence.

Q:

You were at MIT for one year.

Yoshiaki Shirai:

Yes.

Q:

Did you mainly focus on this project?

Yoshiaki Shirai:

After MIT?

Q:

At MIT?

Yoshiaki Shirai:

Yes. I worked alone, and in autumn, my first version succeeded. Then, I asked Minsky to come and look. He looked at me. He admired me, and then he said, “Please come to my house for dinner.”

Q:

Oh, good.

Yoshiaki Shirai:

I was surprised. I rode a bicycle, so I just followed his car, a Ford. I met [Patrick H.] Winston who was also interested in vision. He was also impressed. Since then, I have a very good relationship with Marvin Minsky, Pat Winston, and other members. Until that time, I was alone.

Q:

They were interested in working with you.

Yoshiaki Shirai:

Oh, yes. They asked me to stay longer, but at that time, there was a new project in Japan, a big project. The pattern information processing project started the next year. The main person in that project in my laboratory, ETL, called me directly to come back to contribute to that project. My plan didn’t succeed.

Q:

You had to go back to Japan.

Yoshiaki Shirai:

Yes.

Q:

How was Minsky when you went to dinner or talked to him? How was he to work with or talk to? What kind of things did you discuss, or what kinds of things were interesting?

Yoshiaki Shirai:

I don’t think anything was interesting.

Q:

Why is that?

Yoshiaki Shirai:

Sometimes we talked about robotics. For example, I said the human being is quite different from robotics. I said, “We point somewhere not by calculation, but by memory,” and he understood. Similar to vision, human beings look at things and interpret things, but it’s not only by the line drawing, not with the stereo, but with much knowledge about the thing.

Q:

Right. Yesterday, you mentioned that when you were giving the talk, you showed the pictures; experience of the world and the body. So, you were already thinking about those things in the 1970s.

Yoshiaki Shirai:

Minsky’s way of thinking is very close. Minsky just denies the perceptron very theoretically, but I think experimentally. While studying for my master’s degree, I read his paper, “Steps toward Artificial Intelligence,” and <inaudible 00:29:19> was much impressed by that journal article. I read many articles, but I remember that article very well. Minsky told me that he had much interest in what I said, so I was very happy.

Q:

Did you use this kind of thinking about human perception in developing any of your vision projects?

Yoshiaki Shirai:

Ahh, human vision. We didn’t know if that was human vision or not, but when we reflected myself by the frontal part of my face and brain, we had the option to interpret the results from the line drawing as such, or we could make the line drawing as this. We don’t -- I don’t care for that mechanism of the brain itself.

Q:

It was more just understanding the functions, and then interpreting for the machine, rather than trying to replicate human cognition?

Yoshiaki Shirai:

Yes.

Big Pattern Information Processing Project

Q:

When you went back to Japan, what was the big pattern information processing project?

Yoshiaki Shirai:

The big pattern consists of the 3D image, image of 3D, 2D character, and speech. I was in charge of the 3D object.

Q:

Basically, this was for reading kanji or something.

Yoshiaki Shirai:

Yes, kanji. The big project -- the contractor consisted of many industries, companies, and laboratories because it’s a MITI project -- Ministry of International Trade and Industry. We are <inaudible 00:31:23> the MITI [Ministry of International Trade and Industry], and the project included our MITI laboratories and many industries.

Q:

What were some of the industries that were involved?

Yoshiaki Shirai:

Most were famous electric companies like Hitachi, NEC, Toshiba, Mitsubishi, and Oki --.

Q:

Why were they interested in this system? Why did they want to make this system of pattern recognition?

Yoshiaki Shirai:

Oh, that project was established while I was at MIT. So I didn’t –

Q:

You’re not sure why, but was there an application?

Yoshiaki Shirai:

I was asked to come back because I was away at this time.

Q:

Where they developing an application?

Yoshiaki Shirai:

Yes.

Q:

What was the application?

Yoshiaki Shirai:

The application was for speech and character recognition; that’s clear. For 2D imaging, it is automatic analysis one, maybe, map and maybe the human face like this. For 3D, it is similar to robotic vision, so maybe object <inaudible 00:33:05>.

The object for handling has no direct application, but if it is successful, it can be used for many areas. The project itself has no direct application. The technology is most important. There was criticism because much money was spent, but there was no product. However, the company is very glad to develop their own technologies and utilize them to make the products.

Q:

This kind of technology could go into the industrial robots, manufacturing or something like that. Did you work on all parts?

Yoshiaki Shirai:

Only 3D.

Q:

Only the 3D. What were some of the interesting problems that you had to deal with?

Yoshiaki Shirai:

At that time in my laboratory, I thought that we would have to work mainly on two things. One is to take a picture of the 3D object and understand the scene. The other is to take 3D data using rangefinder and interpret it.

When I returned to Japan, there were researchers looking for themes, so I gave that theme to this researcher, and he worked on rangefinder.

But, I myself cannot work <inaudible 00:34:46>, but I work better together. No one is working on these pictures, so I changed my mind to do self-recognition of a desk scene, which was a usual desk.

Q:

Who was the researcher you worked with on that project?

Yoshiaki Shirai:

Oshima. He received the Takayanagi prize. [Kenjiro] Takayanagi is an inventor of the television in Japan. Brown is the inventor in United States, but independently, Takayanagi invented the television, and he became either the vice president or the president of Victor Company. There is also a Takayanagi prize, and Oshima received the Takayanagi prize after I left ETL.

Q:

This is in the 1980s, right? You were working on the picture recognition, so what kinds of questions were you asking? What kinds of approaches were you using?

Yoshiaki Shirai:

At first, I worked on block world, especially at MIT. The background is black and white blocks, but I thought it was not very good because it is a special case. I used just a gray desk, something brighter than gray, and something darker than the desk. Usually, the objects are a telephone and books, a pen, an eraser, a cup, or something like that.

Q:

Generally, this was just recognizing the object. Did this relate to any work still being done by the robot or by the manipulation group?

Yoshiaki Shirai:

I don’t know, because I developed how to find the edge of the curved object and how to the represent the curve by a second <inaudible 00:37:18> -- curve. It’s not easy.

That kind of basic idea is used, but because there are especially no applications for analyzing the desk scene. Because I belonged to the laboratory, I started from block world to the usual object. Later on, I realized that it was not so good because it was very difficult. It takes five years. Even if it succeeded, there are no direct applications. It should have worked better.

Q:

But now HRP-2 can pick up the cup from the desk.

Yoshiaki Shirai:

Yes. The 3D group consists of me and co-workers, including Oshima. We worked on the analysis and the recognition of the mechanical parts. We worked mainly on the parts of the automobile, stuck it together and – tried to recognize the result.

Q:

So more closely can be applied to the automotive industry.

Yoshiaki Shirai:

That was successful. That’s why he received the Takayanagi Prize.

Q:

Oh, that’s the project he received the prize for- Q2: That’s rangefinder, right?

Yoshiaki Shirai:

He also received the best paper award from the Society of Electronic Communication – Institute of Electronic Communication. That was successful. I don’t think that the method is fairly new. It may be new if good people someday think of this, but fortunately, we have the rangefinder and good data. We could start earlier, so that’s why we can make very good work. <inaudible 00:39:35> Lucky.

Automatic Selection and Imaging

Q:

So in the beginning, when you said it took five years working with the desk object recognition. What other projects did you do after that?

Yoshiaki Shirai:

After that, I worked on automatic selection – a screening of a stomach x-ray picture, which gave an X-ray image of the stomach. This turned out to be very difficult. At the time, chest analysis was already famous, and that is of course because the many organs that overlap caused the difficulty. When you take an x-ray picture, we drink barium, allowing us to get the auto line a little bit easier. The problem is that the shape of the stomach is different from person to person. In addition, the thickness or the consistency of the barium is different, so the contrast is not always clear. Sometimes, the barium goes to other organs, meaning that the boundaries are not clear. This is a problem, so we worked with a doctor in Tokyo University.

Q:

How did this project start? Do you know? How did ETL become interested?

Yoshiaki Shirai:

No, I started myself. ETL at that time was very free. Here, I thought I would like to work on that.

Q:

How did you find out about this problem?

Yoshiaki Shirai:

I thought that the vision was a little easy to obtain. I wanted to find a very difficult one, which cannot be completed for a long time, and then I selected the project. Of course, during that time, I was acquainted with a researcher in the hospital who was working on the CT [computed tomography] of the brain. I worked with her to find the main part of the brain using the CT <inaudible 00:42:36> how to make the 3D volume like this. However, that is not the focus of my work.

Q:

What year was this?

Yoshiaki Shirai:

CT was two years, and for the stomach, many years. In the middle, I became the director, and then I redirected this work to the successor. [Laughter.] So difficult. I wrote papers, and we collaborated with a doctor in Tsukuba University - A medical course because we moved from Tokyo to Tsukuba.

Q:

Did it go to AIST [Advanced Industrial Science and Technology]?

Yoshiaki Shirai:

No, our laboratory moved. ETL moved from Tokyo to Tsukuba in 1979. After that, it was far from Tokyo then <inaudible 00:43:55>. That doctor wrote his own papers. We took many papers, but the useful system was not yet built, unfortunately.

Q:

Why was it not built?

Yoshiaki Shirai:

It costed too much, and we couldn’t expect a special case. Mostly, it was cost.

Q:

What kinds of places were you publishing papers then?

Yoshiaki Shirai:

Information related societies or electronics related societies such as the Information Society of Japan or the Institute of Electronic Communication. Recently, it changed its name -- Electro -- Information and Communication.

Q:

Even longer.

Yoshiaki Shirai:

It included Information.

Q:

But was there a lot of communication with researchers from other countries.

Yoshiaki Shirai:

Sorry, it was the Institute of Electronics, Information and Communication. It abbreviated as IEICE. [NOTE: IEICE actually stands for Institute of Electronics, Information and Communication Engineers.]

Q:

IEICE. I’ve got it.

Yoshiaki Shirai:

It is one of the biggest societies in Japan.

Q:

Did you mostly publish in Japanese circles or also outside of the country?

Yoshiaki Shirai:

Mostly Japanese and I contributed to an international conference. Very rarely did I publish in English.

Q:

What kinds of conferences were out there? There was probably no vision.

Yoshiaki Shirai:

Ichikai.

Q:

Ichikai, okay.

Yoshiaki Shirai:

We contributed to Ichikai and ICPR many times.

Q:

Okay, there was a computer vision conference. When did you become director of the ETL?

Yoshiaki Shirai:

I was not the director of ETL. At first, I became the Section Chief in computer vision in 1979. In 1985, I became Director of the Control Division and I gave my work to my successor.

Q:

In the control division, besides vision, what kind of other work was happening?

Yoshiaki Shirai:

It was very large because it included another section, the Manipulator Section.

Q:

Okay, then the robot came back.

Yoshiaki Shirai:

Yes. At that time, I was involved in the advanced robot for hazardous environments.

Q:

What kind of projects were they working on in the division?

Yoshiaki Shirai:

The whole project is very large, and it involved many companies, big companies including Hitachi for locomotion, Mitsubishi Heavy Industry for teleoperation, Toshiba for vision and recognizing the valve or the piping, and MITI laboratories. There are two laboratories, one is ETL, and the other is MEL, Mechanical Engineering Laboratory. At ETL mainly, we worked on planning the assembly of the valve. We also worked on the sensors planning to execute that task, which included the types of sensors that were useful and where we should place it.

Q:

There was a big application. What was the application?

Yoshiaki Shirai:

It said maintenance in the nuclear plant because the name is for the advanced robot for hazardous environments. However, we don’t think that it’s inside a plant. If it were inside a plant, we would need to think of the radioactivity. We must also think of the camera, which was a bit different, but we don’t think it’s in there. That is a normal plant. There is no difference. We have an interest in robot technology, not nuclear. [Laughter.]

Q:

So, it wasn’t really put in the field. It was still more of a lab project.

Yoshiaki Shirai:

Most of the research, but not only the research in national laboratories but also in the companies. The name is for a nuclear plant, but it’s actually a factory. That gap had some problems later.

Q:

Why did they become interested in nuclear plants?

Yoshiaki Shirai:

Hazardous conditions consist of the nuclear plant being underwater and the fire for the petroleum plant. The fire for our technology has no relation. It’s quite different from the industrial robot. In addition, there is a company underwater, the Mitsubishi ship company for example, making a ship that can move very freely. That is not useful for our industry, so we have an interest in nuclear plants since it’s similar to usual plants.

Q:

In terms of the technology, what were some of the challenges?

Yoshiaki Shirai:

In my laboratory, there was a very good researcher, Katsushi Ikeuchi. He also worked on the recognition of 3D objects. It’s stacking <inaudible 00:51:17> each other using stereo, or what he called a Photometric Stereo. When I was the chief of computer vision, he entered. He was mostly focused on our work. I was supervising him, and he left to go to MIT, came back again, then left for CMU [Carnegie Mellon University. There was also some other guy we called Sugihara. He entered here after finishing his master’s course, and I gave him the theme used on my rangefinder to make the line drawing. Using the rangefinder became much easier, but it was a very complicated scene and very effective. The idea is similar to what I did before the line drawing; just picture <inaudible 00:52:23>. He also got an award for the best paper. We are close – loosely related. Both the ETL boss and researcher for both directors were loosely coupled. When I was a researcher, if I found a very interesting thing, I didn’t ask him if I could work on it. I could’ve done it myself, so I worked freely.

Q:

The funding was very open. Did that change at all?

Yoshiaki Shirai:

The funding? Yes. Funding comes from some projects. Once we get the computer, camera, and some peripheral devices, it can be used anywhere.

Q:

And they didn’t mind in terms of your time? Google lets people use 15 percent of their time on personal projects. Was there something like that? Yoshiaki Shirai:

No restriction.

Q:

No restriction.

Yoshiaki Shirai:

Nothing. If somebody working here wants a hobby, no one complains as long as his work on the main project is good.

Q:

How long did this maintenance project last?

Yoshiaki Shirai:

Oh, I forget. We are working with the task planning like this. For example, [Yuichi] Takase, at the time who was in a manipulator group, developed the direct drive manipulator and performed very complicated tasks. In 1985, Expo was held in Tsukuba, and many visitors came a little before that time. The prime minister of France, [Margaret] Thatcher, and the head of the European Community visited, and we showed them many things. I remember that the direct drive manipulator was broke while the EC directors were looking at it. (Makes noise of machine malfunctioning)

Q:

The curse of the demo.

Yoshiaki Shirai:

It only happened once, so it was a very rare case. However, it happened while very important people were looking at it.

Q:

Bad luck. Yoshiaki Shirai:

Yes, bad luck.

Q:

So, did you work –

Yoshiaki Shirai:

[François] Mitterrand, the French Prime Minister, visited with Thatcher. When Thatcher came, we gave her the portrait of her face.

Q:

From the robot’s point of view?

Yoshiaki Shirai:

Yes. The vision group made it.

Q:

Did you still work closely with the manipulation group?

Yoshiaki Shirai:

Still – what do you mean still? At that time, I was director.

Q:

I’m trying to understand how the groups worked together and if they worked together a lot.

Yoshiaki Shirai:

We did not work together a lot. Sometimes we worked together. I told you about vision feedback. We had to demonstrate some new things, and Professor [Yoshiko] Tsuji as well as Inoue’s boss asked us to perform visual feedback. Then, we cooperated with each other.

Q:

Which year was this in?

Yoshiaki Shirai:

It was in 1970 just after he entered ETL. I wrote a paper in English, received comments, and then had to rewrite it. I was at MIT, and I asked Minsky to check my English. [Laughter.] I remember.

Q:

When did the plant project end?

Yoshiaki Shirai:

It lasted about seven years, so 1998? I don’t remember well. I’m not sure if it’s right.

Q:

Was it in the 1990s? Oh wait, well, it’s during Thatcher’s time, so –

Yoshiaki Shirai:

Yes. If I see this and look at it on the computer, yes.

Q:

Okay, it’s no problem. How long did you stay in ETL? Until 1988, right?

Yoshiaki Shirai:

Yes.

Q:

Did you do any more projects at ETL while you were there?

Yoshiaki Shirai:

Myself?

Q:

Or with the group?

Yoshiaki Shirai:

I did not work on other big projects, but we did small research projects. One was related to robots. I was a little separated from robotics at the time, but I was interested in vision itself. Since I was interested in robotics, I made comments, for example, on the obstacle avoidance of manipulator and sensor planning. I remember that there was a German visitor working on sensor planning, and I discussed and made some comments. They wrote papers, and the Japanese co-workers, including myself, were listed with the author. The German said, “Oh, it’s necessary to include you.” [Laughter.] I remember very well.

Q:

Who was the German? Do you remember?

Yoshiaki Shirai:

Yes, I know. Usually, I remember.

Q:

It’s hard sometimes. If you remember later, you can just email me. It’s no problem.

Yoshiaki Shirai:

Ah, he came from Karlsruhe, Germany, I think. He came back and spent some time, and then he went to Brussels to some European organization as a scientific attaché.

Q:

You kept working on some of these robotics related projects as an advisor.

Yoshiaki Shirai:

There were so many work possibilities. The other section mainly worked on computer hardware. I sometimes discussed that. There were so many directors and so many researchers, so I don’t remember. <inaudible 01:00:55> Oh, Niepold. N-I-E-P-O-L-D.

Q:

How many researchers were in your group? How big was it?

Yoshiaki Shirai:

About forty.

Q:

And when you were a director, did you have to –

Yoshiaki Shirai:

Forty, but that is the number of regular members. We also accepted students and visiting researchers from the companies, so we had more people working in the group.

Q:

That’s a very big group to manage. Did you have to help choose projects or communicate with MITI in anyway about what to do?

Yoshiaki Shirai:

One thing, we are a little famous for in Japan, so many companies come to look at us. Some companies ask us to solve their own problems, so we accept researchers from the companies, and we work together. I worked together with many companies. For example, I have worked with Denso or Matsushita, at that time known as Matsushita Electric Works. We completed some methods and got a patent. That is why there were many researchers. They gave us some problems, and if I thought it was interesting, we worked together using the computers.

Q:

What kind of problems did you choose?

Yoshiaki Shirai:

Of course, I chose vision related problems. For example, the inspection of solder joints or maybe the inspection the surface to see if there were cracks or similar issues.

Q:

Do you want to mention any other project? Did you have any other important projects?

Yoshiaki Shirai:

There were other projects, not my main directorate, but in my division, of course. A robot company wanted to work on planning and force control, so that company’s researchers came and we worked together.

Q:

Which robot company?

Yoshiaki Shirai:

Kobe Seiko and some other, but I don’t remember well, because the work was in the manipulator group.

Q:

They were working more closely.

Yoshiaki Shirai:

I met with them maybe once per year because sometimes I attended the manipulator group meeting at the end of the year. Sometimes the researcher from that company joined the meeting.

Q:

Then they kind of exchanged ideas and maybe brainstormed new projects?

Yoshiaki Shirai:

No. They had an exchange. They discussed what they were doing and current conditions. We would spend very good days together, so we remember that days.

Q:

Why did you choose to leave ETL in 1988?

Yoshiaki Shirai:

I like ETL very much, but the number of the members of ETL is limited.

I think if we stayed on, not many people could enter. When I entered ETL on the first day, my boss’s boss told me “don’t think you can stay here for long time. You must quit. You must work very hard so that you can go anywhere.” I always thought it is almost time.

At the time, there was a proposal for Osaka University. Before, there were some proposals, but it was too early.

Then I became the director and I was not directly related to the research work, so I thought that I could not continue because I might be spoiled. It was a very good position. I did not earn much money, but the work was very good. I thought I couldn’t continue for a long time.

Q:

How was the university different from working at ETL?

Yoshiaki Shirai:

Not free universities. There was a lack of money. The most important thing is that at ETL there are many researchers in the same field. However, in universities, they are still mostly professors, associate professors, assistant researchers and students. This is called a “vertical relation”. In ETL, it’s known as “horizontal relations,” it means equal, so we can discuss freely. In this lab, the discussion is not so equal. That’s the big difference.

Q:

Did you enter as an assistant professor or higher?

Yoshiaki Shirai:

Yes, as a professor.

Q:

As a full professor.

Yoshiaki Shirai:

Yes. If I entered as an associate professor, maybe I could not become a full professor.

Q:

Were there people at the university doing computer vision before you?

Yoshiaki Shirai:

No. I entered in mechanical engineering, but it is called electric control. It’s closer to information, but basically, it is mechanical engineering. No one was working on vision.

Q:

You were the first?

Yoshiaki Shirai:

Yes.

Q:

How did you develop?

Yoshiaki Shirai:

I was the first one in that course. However, there were some other parts that were working on vision.

Q:

Did you work with them?

Yoshiaki Shirai:

Yes, in laboratories in an electronic course. There were some special laboratories. In some laboratories, they worked on vision.

New Research in Assistive Robots for Humans

Q:

How did you start your lab? What students and money?

Yoshiaki Shirai:

I decided not to continue my research in ETL, so my lab was quite different. First, we had to prepare the basic tool for students not researchers. You start from a very easy scene -- so the recognition of the outdoor scene or stereovision for the indoor scene because stereovision is not so quick. A rangefinder is suited for industrial applications. Stereovision at that time is not but is still not far from the real application. Stereovision and outdoor scene understanding are not useful for industry and motion picture analysis, for tracking human beings.

Q:

Why did you decide to do that? Why change?

Yoshiaki Shirai:

Yes, later – of course I was working from a robot, which I will later mention. At first, the concept is that if I work, we compete with ETL. It’s not so good because we know each other. That is also good for industry, and that means it is not good for students, which is the motivation. At first, I was working with Professor Minoru Asada, he’s also famous, and he moved to another laboratory as a professor. I then worked with Yoshinori Kuno, who spent one or two years in CMU and came back to Toshiba. We worked together in a big project.

I know he’s very good, so I asked him to come. He’s interested in human beings, so the interface with human beings, and even in his philosophy, is not possible for the computer to recognize everything. If we don’t know, the computer doesn’t know. We asked human beings, which I think is a very good idea.

After he left, I decided to work on a similar one, which brings things for a handicapped person or elderly persons. That is very useful because at that time, my father, although he already died, sometimes asked me to bring him something. People say, “Oh, for the meal can only be one time. For meals, human beings can come and eat, and that’s okay, but please bring me something.” It must be prepared every time this happens, so I think that’s a very good application. Then, at that time, there was already a demonstration. For example, in Tsukuba Expo 1985, there was a demonstration by Tokyo University. One person was in pain, and a robot came to take the bottle on the table and bring it to them. I thought usually the bottle was not on the table, but was instead in the refrigerator.

The difficulty is how to pick up the bottle from the refrigerator. Instead of, how do you put this – <inaudible01:13:29>. Then, take some drinks from refrigerator, go and deliver. I started this kind of work.

I did part of vision. At that time, my assistant professor was working on motion for an automatic vehicle. He is now a professor at the university where he is working. Now, he is famous in Japan. When the robot cannot understand what the object is, he asks which object it is. It notifies the person that it can’t recognize the object they ordered. It then sends the picture to him, and the person can say, “Just behind the Coke.” The robot would understand that the item the person would like is just behind the Coke. Interaction is also useful, so our laboratory includes speech recognition and teaches how to interact. Speech recognition itself is not my work. It is our work and a way of interacting that is precise and quick.

Q:

Did you start working on this immediately in the late 1980s or was this in the 1990s?

Yoshiaki Shirai:

I started in 1988 from the very basic themes, the outdoor scene, stereovision, tracking people, and then the robot that bring things.

Q:

Then you actually had to do some studies with people.

Yoshiaki Shirai:

Yes.

Q:

It is kind of like a human robot interaction study. You are developing some of the interface?

Yoshiaki Shirai:

No, it was just at that time of interaction. For example, if a robot is certain that something is the desired object, it tells the person, and “I think it is this object. It is this object, or is it this object?” <taps table> “Is that okay?” Then the person can say, “Oh, another one.” If it is unclear, then it tells the person that it found certain objects, but it doesn’t think it is the one that they ordered. The person then says, “Oh, next to this object. That is easier.” If without the saying, “Oh, I cannot find the object itself, what shall I do?” Then the person <inaudible01:16:54>, because it does not know.

Q:

Right, how to explain which object?

Yoshiaki Shirai:

Yes. Similar to asking what is the easiest. It also extends the object from the container of fruits, many kinds of fruits. The robot shows the person, and the person decides if they want that fruit. They’re seeing an object. We’re working on how to profile good information that is easy to specify the object that person who orders would like.

Q:

You mentioned you were working with somebody who had a robot who was implementing –

Yoshiaki Shirai:

Yes, in my laboratory, I had a robot.

Q:

It was a little mobile. What kind of robot?

Yoshiaki Shirai:

Yes, it should be a vehicle because it must open the refrigerator.

Q:

It was a mobile robot with an arm?

Yoshiaki Shirai:

Yes. It moves in front of the refrigerator to open it and take something. It requires a rather large arm, so it is a heavy, large vehicle. [Laughter.] On the manipulator itself, we did not make much progress. We follow the current technology.

Q:

What other projects did you do in your lab with your students?

Yoshiaki Shirai:

We worked on its application of human tracking. That is, the analysis of a soccer game. The player and the ball. You ask about that goal scene <inaudible01:18:46> -- and the success of the long pass. Then it shows the desk scene.

Q:

Do you want to mention any other projects?

Yoshiaki Shirai:

Yes, many projects. [Laughter.]

Q:

Mention whichever ones you think are important.

Yoshiaki Shirai:

Many people are working in gesture research. Gesture is very easy, I think, because it cannot express so many things. However, do you know that sign language for deafness is very difficult because it overlaps with each other and with the face? I found it very difficult. We moved to recognition with the handshake.

One person, working with his students, found how to find 3D structure from the contents. Now, he is with me at Ritsumeikan University. There are a few Ph.D. students working on how to find the 3D shape of human face hand posture by overlapping with each other or <inaudible01:20:19> background. Some parts are very similar to the current <inaudible01:20:27>. This is one of the big research subjects. I might have forgotten one other subject. [Laughter.]

Q:

Since you have been in the field for a very long time, how do you think the challenges of computer vision has changed through the years?

Yoshiaki Shirai:

Oh, in general?

Q:

Yes.

Yoshiaki Shirai:

In general, I think there are two things: one is that the information, which can we use increases the monochrome, color, and motion pictures. The motion picture comes from many cameras similar to this. Another change is that the speed of computers has increased. Memory increased, and hardware changed. Now, we can use methods that could not have been used in the old days, especially for motion picture analysis. You can use it online.

Q:

What about the connection between computer vision and robotics? In your work, having the robot do something helped come up with problems for computers. How do you think those are connected?

Yoshiaki Shirai:

Manipulation and motion – locomotion itself is developing, and the vision is developing. At first, they’re in primitive stage and they’re working together, but as they develop, it’s not easy to work together. The current situation is that for manipulation group or locomotion group they make use of the current technology, and for us, the vision group, we make the vison. Sometimes, we need a manipulator for locomotion, so we use the current technology. In the old days, both are advanced together.

Q:

So, it’s not as easy anymore to work together.

Yoshiaki Shirai:

Yes. Sometimes, we integrate it, but at that integration, we don’t need both teams. One team makes the importance of the vision group and would like to integrate to ask the manipulator <inaudible01:23:01> like this, vice versa.

Q:

What about in your group – in your work. What are some of the challenges you are interested in now? How have the kind of questions and challenges changed?

Yoshiaki Shirai:

I left and went to Ritsumeikan University about six years ago. I continue hand gesture analysis and service robot research. It still gets difficult.

Q:

After Ritsumeikan University, where are you now?

Yoshiaki Shirai:

Now, I’m in the last stage at Ritsumeikan University.

Q:

Oh. Last. [Laughter.]

Yoshiaki Shirai:

I’m still a professor at Ritsumeikan University. I’ve been a professor there for six years.

Q:

And at Ritsumeikan, is the funding situation still the same as in Osaka kind of difficult, or where do you find funding for some of your work?

Yoshiaki Shirai:

When I moved to Ritsumeikan University, I had lots of money, ten million Yen. We could freely use it, so we bought manipulator and computer mainframes. Unlike Osaka University, funding is not a big a problem because we had lots of money. We have not suffered from funding problems.

Q:

Osaka University has a lot of robotics activity happening, especially with some of the humanoids and androids and all kinds of –

Yoshiaki Shirai:

Yes.

Q:

Yes, Ishiguro Sensei. [Laughter.] Did you have any connection with some of those people?

Yoshiaki Shirai:

In research, no. However [Hiroshi] Ishiguro – [Minouri] Asada is still <inaudible01:24:56> [Saburo] Tsuji, Professor Tsuji, and I know he is still -- and Ishiguro is still <inaudible01:25:02> Tsuji, so while he was student, we really knew each other. They moved to Osaka University, my campus.

They went to another campus, but he moved to this campus. We know each other. We discuss very often, especially with Asada once we have a team, so --. After we separated, we had no direct relation in the research. For example, Asada has a research interest in how the intelligence of a baby starts. Ishiguro has an interest in human beings themselves, and for him the robot is used as a tool to analyze human beings. We are interested in the robot that is useful for human beings, but I think we have different interests.

Q:

What are some or the challenges you think in computer vision coming up, or robotics?

Yoshiaki Shirai:

In my laboratory?

Q:

In your lab and then in general?

Yoshiaki Shirai:

Oh.

Q:

But first in your lab.

Yoshiaki Shirai:

In my lab. Yes, I continue, and another one is to work with the analysis of human motion with just one camera. We started resampling, especially for golf swings. We started when I was in Osaka University working together with a company that produced golf equipment. Then, one system we completed consisted of two indoor cameras. It’s easy because we could take background images, and when a human being approaches, we could subtract. It’s now more difficult with only one camera and without background image.

Q:

So, outside basically.

Yoshiaki Shirai:

Yes, outside. Daytime and night. It’s very difficult, so we are working on it.

Q:

So, this is to kind of help people learn how to do the right golf move?

Yoshiaki Shirai:

Yes. Advise -- You are right; this part is wrong. My coworker, Shimada, once worked on this, and is still working very quickly using many computers <inaudible01:27:38> very precise also <inaudible01:27:44>. Most of the work is for gesture, which is easy since its discrete hand sign language is a little difficult because there are many kinds of shapes. Shimada has an interest in it without any application. Very small differences can <inaudible01:28:10>. They may be more about medium kinds of shape. [Laughter] That is also a challenge because it’s a 3D shape.

Q:

And for computer vision, in general, what do you think some of the future problems are?

Yoshiaki Shirai:

Before that, I used many pictures collected from the Internet as training, and I tried to recognize all of it. That approach is so popular that I avoid it.

Three years ago, some students from Vietnam came here for their master’s course, tried to work that thing, but there was no one who could take care of him. Because of this, I have to take care of him. We work together and make a system that can recognize a car, bicycle, airplane, horse, [laughter] and sheep like this using samples. I found that it’s possible to a certain extent. Maybe 90 percent is successful, but the problem is that if it is not successful, we don’t know how to change the algorithm. Because we don’t analyze logically, it just gives many examples, so that’s very popular. I continue because it’s easy, because there are many selected features of images. Using that features for giving <inaudible01:30:56> select the many features. It determines what kind of features are available and extracted. This is a horse. Another horse is coming, so there are many pictures. If this is a not a horse, it scribbles in an automatic way, then there are some systems that mostly correctly can recognize the horse. That is a little different from human beings. It is not the tendency, but many people realized that is because recently in Japan, the Japan Information Society had a special issue on vision. They asked what happens if it is not done, and it said to only use one picture. Without many retrieving pictures is how to understand the picture. That is the challenge. That is what I thought at first. I used only the pictures. The difficulty was not easy, so I think that person who knows very well could track vision. Not many people would track because it’s difficult; it’s not easy to write papers. The current technology for machine learning is very popular. They succeeded in face recognition because they have many samples of faces, and many researchers work together. Face recognition <inaudible01:32:17>. If the killer application is found, then that vision system will be made in a few years.

Q:

But nobody knows the killer application quite yet.

Yoshiaki Shirai:

Without the killer application, they cannot concentrate on the program.

Q:

We have one final question that we ask everyone. It is kind of for education. If you could talk to some young students who are interested in robotics or who are interested in computer vision, what would tell them as advice for how to start?

Yoshiaki Shirai:

At first, they must have significant interest. If it solves the problem, then they have no interest. I was a young student at that time that Eiichi Goto, who is very famous for the Parametron Computer, was interested in vision. He said that he is not working on vision; he’s a physicist. There are many pictures taken in a physical experiment, and to analyze them, it takes a year. To take pictures, it takes maybe a week. He asked if I could help by analyzing because the vision people were analyzing them. I listened to this. Oh, vision is very useful. It is not solved, that’s very important. We must not tell them that this is complete. This is very interesting. Then, it’s not interesting. [Laughter.]

This is not solved. This is so <inaudible01:34:16>, he asked for me to help. That is very important, so students also try to find what is lucky. If it is helpful, they want us to come in and analyze, but not just apply the current tool and the conventional way with just small different applications. Everyone can do that, so that is interesting.

I graduated from the field of mechanical engineering and moved to vision. That’s why I can do many things, which cannot be done in other fields. Robotics includes many fields, so even if they think I am not related to robotics, it isn’t a problem. Someone comes to the robotic field since there are many things to do.

Q:

Thank you. Is there anything you wanted to add that we didn’t cover?

Yoshiaki Shirai:

I don’t exactly know the purpose of this interview; is it the history?

Q:

Yes.

Yoshiaki Shirai:

Around 1980, in our group in ETL, a hand manipulation had fingers working. This was tele advanced. When Mitteral <phonetic 01:36:20> comes, shakes the hand and shows the product, I thought ETL was very advanced in field of robotics and robot carpenters.

Q:

What were they doing?

Yoshiaki Shirai:

Robot carpenters saw the wood and use hammers. Carpenters are the ones who make some object. In block’s world, they said put and <inaudible01:36:59> but in actually make this using the force control. Therefore, I think ETL’s robotic groups advanced in that time – at that time.

Q:

Why did they stop being more advanced?

Yoshiaki Shirai:

One reason is that some person who was very eager left. The other is that once it is demonstrated, they’re done. Many other laboratories can easily follow. We don’t have any secrets. We very open, so --.

Q:

I am curious; you said you worked with Masahiro Mori. How is he to work with? I know you said you didn’t see him very often.

Yoshiaki Shirai:

Oh.

Q:

He’s also very famous.

Yoshiaki Shirai:

Yes, he is.

Q:

Young kind of.

Yoshiaki Shirai:

He likes the peculiar things. I’m a little orthodox. For example, he gave one lecture for us. He only gave one lecture because he belongs to the laboratory, not the school. I attended, wrote a report, but the evaluation was not so good. I asked him why my evaluation was not so good, not the best one, the second. I think I have the best score in my field, so --. He said, “you are the like a sample.” “Nothing’s special,” he said. “So, you should have said something before that.” [Laughter.]

He accepts everything, but he doesn’t teach. That’s the way they teach, but usually when we enter school, the course, maybe the professor gives him types of things or asks if he likes some of them.

When he came, he said, “This is the laboratory, not the university. No university here, this is the laboratory. So, you are not the customer.”

Most of <inaudible01:39:48> industry.

At the time from the industry, you could use money. You couldn’t use much money, and you must select the theme. I made an electric simulator when I was working.

And one day, <inaudible01:40:17> he came here. Oh, you have made these kinds of things. Soldering is very bad. You pench [Note: “pench” is a Japanese word that means “plier.”], you cut, cut, cut, cut. You mustn’t do it again. He left. [Laughter.]

It is very strict, but he knows that I can continue even if --. So, someone is working on something mechanical. <inaudible01:40:54>, it is not a success if the professor says nothing. Very good days, even in situations where someone is good, someone is bad, but not all bad. At the time, Masahiro Mori was very young, but after that, he got a little older in Tokyo Institute of Technology. I later met him. He’s a gentleman.

He changed very much because it’s a school, and he must teach students. These are undergraduate students. He knows what the students are. [Laughter.] He’s very creative, and he started a new project, a robotics project. It is a contribution.

Q:

I heard you also had a research institute for [Yusuke] Mukuta or something.

Yoshiaki Shirai:

It’s a research institute for manufacturing, yes.

Q:

It is kind of like Buddhist principles of manufacturing.

Yoshiaki Shirai:

Yes. It is a manufacturing laboratory like this. He has an interest in Buddha. He sometimes attends an assembly of Buddhists; not academic conferences. [Laughter.]

Q:

Great, thank you. Those are all our questions.

Yoshiaki Shirai:

Okay.