MIT MechE Symposium: Mechanical Engineering and the Information Age - Masayoshi Tomizujka, E. Daniel Hirleman, Ian Hunter, Seth Lloyd and Panel

Search transcript...


PRESENTER: You know, I'm a member of our systems and controls division. And we're always being told how we need to catch up to Berkeley there because they have the best division of systems and controls. But I guess the only consolation we can get from is that Professor Tomizuka is, in fact, an MIT graduate. So he'll be talking to us about making products of Y2K status report. Professor Tomizuka.

TOMIZUKA: Thank you very much.

AUDIENCE: That was a bad introduction.


PRESENTER: It makes us look more humble than we actually are.



TOMIZUKA: Yes. Thank you very much for including somebody not from MIT.


On day four. I would take this opportunity to really go through what we are doing in the area of mechatronics at Berkeley. And I'll start with some general comments on the mechatronics. But also yesterday, there was some good talk mentioning about mechatronics. I don't have to, but I have been giving this kind of talk on similar advances,

I just changed the title. I think it's [INAUDIBLE] Y2K saved us before. And if we say mechatronics, there are also mechatronics devices in the market. But given some fine data, I'm not really defining mechatronics in narrow sense, just be part of these product. And in fact, the definition has been evolving. Started in 1969, Japanese started using it. And at the beginning, people really deferred to electromechanical systems.

I was doing the chief editor of the IEEE transactions on mechatronics up until December 1999. There, we used [INAUDIBLE] through the different synergetic integration of mechanical engineering with electronic and intelligent computer control in the design or manufacture of industrial products and processes.

Now if we-- this was in 1996-- but now everybody is talking about information technology. If we would have started journal, maybe instead of electronics, we would have, say, something information technology. And instead of computer control, I would have put more abstract wild-like complex decision making. So therefore, the Y2K definition may need something like the synergetic integration of physical systems with information technology and the complex decision making.

The other part, if you and-- you can still keep, but it's not that important. When we say information technology, I think Nam Suh had some good definition of what is information for information technology, but it's essentially the same thing. I'm putting some devices, which would store information and so forth, but starting from electronics, micro processor in 1970s to some network communication internet embedded DSPs.

Those are home phone, what I mean here by information technology here, and decision making site, I am referring to my favorite area, control theory, design methodology, while in more recent development in the area of hybrid system theory and so forth. So in the mainstream, of course, there is the physical systems. And physical systems, I see gaining one more importance in our life is something like bioengineering and medicine, energy transportation or intelligent transportation systems, environmental engineering, and manufacturing.

I picked this list from a recent publication from ASME. It said, mechanical engineering in the 21st century. And it came out September, 1999. Last year. It's one of a few publications from ASME which I quite agree very much and see quite useful. And by the way, they are listing 10. And of course, at the top, something which would make the difference in the next century, say put information technology.

And the way I see is that the first three more or less talking about some [INAUDIBLE] and everything technology. And merely the last five defining some exciting area. And mechatronics somehow come here to synthesize and utilize a neighboring technology in each specific domain area to make things more exciting.

Now of course, the driving forces of mechatronics was I put information technology with making mechatronics more and more exciting, but suddenly of course, need from industry. And those two combined are introducing new ways of doing things, new patterns of thinking. Some people prefer to call mechatronics simply as a best practice. So certainly that best practice is evolving as the technology develops.

Need from industry. This is from a talk by Mike Masten, Texas Instruments. He put together a general trend in industry. And lots of things already said yesterday. And some other thing which may end up having explicitly said, but I think you know most of the things here. And one thing probably, which interesting is this teams-- team works are preferred approach.

So I would comment later about some of the courses that we teach in this area, we try to teach some teamwork through the project work. Now let me go through some decent research at UCB in the area of controls of mechatronics. And good part of the topics that myself and my control colleagues. But I added something which is really relevant to information technology, which is not quite my area, but I cite work by Professor [INAUDIBLE] and some of the [INAUDIBLE]'s work. I think she's a member of the Advisory Committee and she'll be coming tomorrow.

Now the area of computer disk drive-- this research is conducted in the computer mechanics laboratory. And it's a consortium of industrial consortium. And in the controls area, myself and Professor Horowitz have been looking into this area. And really to answer to the need from the hard disk industries. They are striving for higher density storage, faster seek time, and of course, lowering cost.

If we read across some schematics what's going on, there are two types of problem. And one is so-called track following-- the decoding head has to stay on the data track. And I think it's supposed to be circle, but it's not quite. There is some random error, while in case of removable storage devices, there might be some eccentricity. And there are also a lot going on, but some decent work is one of my students finishing in the area so-called repetitive control.

This is a methodology which vary data through its periodic disturbances. And repetitive control is ideally suited for this type of problem because disk is spinning at some fixed speed, so the same pattern appears many times. So it's essentially how you may drop sensitivity to almost to 0 at certain frequencies.

The disk drive industry use so-called sector drive. Sector drive is that the fact that this can be sliced into the area and so-called provisional information is not written only at the boundary. So you would like to keep a number of sectors small so you have more area for storing data. But sometimes that you cannot get enough-- the frequencies, medium end frequency for doing some fine control.

Now instead of adding the sectors and putting more information, there has been some work to increase some effective bandwidth by updating the input more frequent than the measurement frequencies. And that's called multi data control. And it's a very interesting area, both from control point of view and for this type of application point of view.

And another student is finishing her PhD-- in this case, she-- in this area. And also the traditional disk files that you find in your computer has got only one actuator [INAUDIBLE] motor, but everybody is now working on how you can introduce second stage actuator. Essentially you put more freedom. So for fine motion control. And the next there is immediately coming out in the market is on utilizing piezoelectric piezo transducer. And certainly we are working in this area.

Also my college Horowitz in collaboration with Pisano, they are really MEMS faculty member is looking into the next generation second stage actuator by utilizing MEMS technology. And this work is sponsored by DARPA. And they are working with IBM. So this is their rotary type actuator. They designed both rotary type actuator and the linear actuator, two point millimeter diameter, and they have already closed loop. And they have got the basic performance.

One area I'm working with Horowitz is an NSF GOALI project with Xerox corporation for making the paper handling part of the copying machine more desirable and more intelligent. And it's very a mechatronics product. And we are suggesting to introduce some mechatronics syncing in the product design. So we are looking into some new ways of handling the papers in the copying machine.

Traditional machine, essentially you can think about that there has got one big conveyor belt. Paper is fed from one end and will do the image transfer station. And there is some control going on, but from here to there, there is essentially no control, open loop control. So as a result, the problem you experience is so-called soft jam or hard jams, when some measurement station, they are checking whether the paper is present or not.

If the paper does not come at the expected moment, the computer, the controller, thinks that something went wrong and simply stops the machine. And even though the paper may not be jamming, but software found the jam, so it called software jam. And we are trying to make it better, at least remove software jam, by breaking up the conveyor, big conveyor, into several sections and add sensor and actuator at each stage.

Certainly, this will complicate the problem. And if you have many sensors and actuators, if you do not have a good way of deciding the control action, this is not the kind of scheme that you would like to employ. But with the power of the computer now, this is quite feasible. And so we are looking into this problem.

Still, we have to do some economic considerations, so what is the required number of actuator and sensors? We can not just add too many from cost point of view also, and also reliability and maintenance. And also the problem itself is quite interesting because paper is moving and there are lots of constraints generated when paper is handed from one station to another.

You do not want to back up or you do not want to stretch because you might end up adding some marks. And also if your purpose is to essentially control the distance from one paper to another-- so if it's too close, we would like to increase the separation. But there is only a limited time window that you can apply the control, feedback control.

So it turns out to be a nice example of the hybrid system, discrete event systems. And intellectually, it is very quite challenging. Most of the thing we do, we would like to do experimental check. Otherwise, people won't believe us. But also by working on the hardware, we find lots of interesting problems.

So in this case, we got part from Xerox and we made some desktop paper pass. And it has got essentially three parts. Each part can be independently controlled. And of course, we have some paper feeder. And paper can ease of going around for many times while we have some exit gate that controls so paper can come out.

This is a motor. And measurement at this stage is a motor measurement. And also the roller is very physically moving the paper. But we are measuring the position of the roller. It's best to measure where paper is, and also if we can measure the paper velocity. So we are also adding photo sensor to detect the position and velocity of the paper.

California has got a PATH program. It's a very big program. PATH stand for Partners for Advanced Transit and Highways. It started in 1986. And there are lots of interesting control programs in the PATH area. In particular, advanced vehicle control and safety systems, it's lots of interesting control problems for mechanical engineering student, mechanical engineers.

Professor Hedrick, a former faculty at MIT, and myself have been working a lot in this area. And especially from automated driving, AHS point of view. In 1997, we had a high point of the past work. This was a slide from the 1997 National Automated Highway Systems Consortium demo down in San Diego.

And essentially past work, the research work has past resulted into the demo by eight car platoon. So eight car is making some sort of chain operation. They have very tight longitudinal control. And Professor Hedrick's work really resulted in doing this tight longitudinal control. My group has been working on automated steering for many years. And that really helped doing this demo.

Essentially what we do is we put magnet in the middle of the road. Magnet is about four feet apart. And magnetometer sense distance and then steering action is figured out by computer. So it's completely hand-free driving. He's not putting hand. And now after this demo, of course, the National Consortium was disbanded, discontinued. But now California is continuing this automated highway effort with emphasis on these heavy vehicles.

So I'm currently working with my graduate student on the automated steering for this track. This is flatline track. In fact, this is our current test track. So magnet technology has being shifted, applied, to some of the maintenance, load maintenance work by Cal Trans people-- California Department of Transportation. We have some small countries and they have got the idea to put magnet and do some guide vehicle while for the maintenance operation, for safe maintenance operation.

They are not immediately interested in automating the operation, but try to come up with some way of assisting the human driver. When you removing the snow, when the wind blows, there is so-called white out conditions generated, thereby instantaneously you lose sight. So in that case, how is magnet information combined with some dynamic prediction of where the vehicle may be going may be displayed on the screen so the driver can easily detect-- why, he even can drive looking at the screen.

It's an interesting man machine interface design problem because you'd like to see that driver be not confused by looking at screen. He should be able to switch back and forth and still maintain some semblance of operation. And one student is finishing PhD currently on this subject. So this is a typical application. This is not snow plow. This is a snow blower.

And in case of snow blower, Caltrans are interested in control even with automated driving. So we are starting that project. The Kazerooni, Professor Kazerooni, he's a graduate of MIT also. He has been working on a combination of the human intelligence and the power available from the hardware devices.

And he has been coding extenders. He joined us from Minnesota probably five, six years ago. And this is a device he designed at Minnesota. But it's a very early version of really making his idea of human intelligence plus mechanical power into some hardware device. And more recently, he has been working on-- from arm to leg. And this is some of his recent work. Power blower extremity enhancer.

And you can see some device. And this is a recent experiment he conducted at Berkeley. And the idea is by having this device, people can carry lots of heavy things and some battleground and so forth. By the way, this is one slide he requested me not to put on the web. So I will just eliminate this slide when I give you a copy.

This is the Professor Wright et al. Sanjay can explain this slide much better than I do. But this is light has some interesting idea. Now everybody is talking about how you can make use of the resource available at remote locations to accomplish your job. So efficiently networked manufacturing service. And this area is very-- this activity is expanding. I think Ford Motor Company has donated quite a bit money to even try this kind of thing on the education side, education aspect.

Now this is a slide that I received from Alice Agogino. So I cannot add too much, except it is written down here. This is her research project, the concept database, an overview. It's essentially a multimedia network conceptual design support tool. And some smart navigation through a hypermedia database uplink design concept and electronic component catalogs.

And her present research address aspect this there here. Some of the things which I personally find interesting is to manage uncertainties in design decision making, intelligent real time design. So how real time-- this network based tool can assist the designer. She has made some prototype conceptual database.

And in fact, one of our goal here in mechatronics is a very important aspect in her research. I didn't quite mention, but at the bottom, she put that domain of application is mechatronics design. So it's highly relevant to mechatronics. And this is for assisting the motor selection. So she put lots of intelligence there.

Now let me just comment some indication of the challenge from mechatronics point of view. And in fact, we are doing at UC Berkeley, as far as I see it, all ME and EE student should be exposed to mechatronics thinking. So all of our student needs to experience some mechanical design information technology and so forth, to team through teamwork.

But also this area, technology keeps changing. So students should be trained to be forward-looking and curious. What we provide at the school will never be the all, and maybe in a few years it's just all this stuff. So unless they have trained to be study themselves or curious look into new areas, we are not doing the right job.

That last comment is more addressed when I give this kind of talk outside the United States. I suppose we are-- everybody is using English. Now, regulatory work is essential in this area. So we have both undergraduate and graduate courses to let students do some laboratory work. Often, we make a team. And through that, people will learn some complexity aspect, the system aspect we are doing.

So several things happens at the same time. Error depends on minor changes in timing. The best way to learn this type of aspect is through actually project work. This is a undergraduate course that Kazerooni is teaching. And so if you look at the syllabus, it's from some microcontroller electronic motors and so forth. So this is covered by the extra.

But on the top of that, he has a project everybody has to accomplish projects. By the way, when student takes this course, student has not taken necessarily feedback control courses. And we can never underestimate the creativity of an undergraduate student. Without control courses, formal control courses, students in fact, does lots of interesting projects.

This is one project from last year-- automatic bicycle gear shifter. It's quite innovative. And some walking robot. And it walks. And to change the direction, it has got another motor spinning. So depending on which way motor is spinning, it's not making a very interesting maneuvering.

This is from across developed by another MIT graduate, in fact. David Auslander. He has been a kind of leader in the area of introducing microcomputers to control [INAUDIBLE]. And this is a switching control. He calls switching control. But essentially it's a discrete event system. Many things happens and how that computer system can detect which operation he's in.

In this particular case, training is going around. At some point, train motion has to be synchronized with the motion of this conveyor, which is carrying some iron ball. Its a magnet. So by on off switch, magnetic can either on this side or magnet can be dropped. Or a ball can be dropped. So it's a loading-- some steel ball on the train while the train is in motion. So just by doing this, the student can lots of the interesting aspects for real-time programming.

Now at a graduate level of education, I have to really say that fundamentals important-- really you emphasize the fundamentals. Fundamentals are really fundamental. And I think we do quite a good job in this area also. Now topics for dissertation research. This is something quite opposite from what Professor Siu. Was mentioning yesterday. Two minutes?



TOMIZUKA: So balance between theory and application, as I say, it's very important. Of course, it doesn't have to be 50/50 split. In my case, I'm quite adjusted to that type of a student. So it's customized. Some student [INAUDIBLE] The theory part could be 90% and application part could be 10%. Or the other way around. Somebody more technology oriented, I could put that and emphasize more this aspect.

But I think at some point in life, especially student life, students should be exposed to both. And of course, that's great to the research funding. But as we have been so far lucky to find some exciting project for student. And especially for PhD. I think at UC Berkeley our goal is to really train future leaders, both in academia and in industry. And so far, we have been quite successful.

When we say fundamental, essentially every aspect-- but if you look at the kind of research courses that we are giving at the graduate level, we are very happy that we have from both technology end and to the theory end. And especially the control theory side, we have a very good coverage and we make it sure that we give this kind of control theory courses every semester.

But it's not very control courses given for the curiosity of only faculty members. We are giving these courses for students. So even when somebody goes to sabbatical, we make it sure that the course is covered. This is another aspect probably of education. This is the outcome of Alice Agogino's effort of national coalition.

Since this coalition is now continuing as a need. National Engineering Education Delivery System Digital Library. And you can check this on the internet. So if you say mechatronics, you can go through some of the modules that she has developed or her group has developed-- not necessarily all from Berkeley.

So let me conclude my talk by stating that information technology and a new dimension to mechatronics. Mechatronics is essential in all of engineering in the information age and lifelong learning aspect is important. And so it's system level thinking and teamwork. Thank you very much.


PRESENTER: Thank you. Yes, Eric?

AUDIENCE: Tomi, you put a lot of stuff in at 25 minutes and talk, but you didn't mention where you guys going.

TOMIZUKA: I think that we are essentially going in the direction. Information technology area, we talked about whether we should make something option at this undergraduate level. There is some divided opinion. We are not quite convinced to make a two pass, because always there's a department did not come agreement. Somebody always wants to keep it the way that we have been doing.

And so then there's some new people would like to do some more emphasis on information technology. So for required courses, we once thought about whether we might be able to make two versions-- one with some computational information technology thing and the other one without. Just traditional way. But I think too large an idea is not a good idea.

So eventually what I am thinking, what I'm seeing, is we will be introducing more and more information technology, at least to use a computational thing in the education of undergraduate education. So that's definitely happening. Now, and also I mentioned yesterday a little bit, but we are at a situation that maybe close to the Silicon Valley. Student attention is very much directed to computer science.

And we have to really make a good case how mechanical engineering will grow in that environment. So one idea, which was the idea discussed among the college was that probably, because our information technology computational things does not have to be taught on the all in computer science, but if we do, whether we can set up an attractive courses that we can draw lots of students.

But at least when we are hiring new people, we are looking into what kind of expertise new guy has so they can naturally participate in the area of computational engineering information technology aspect. So that's one thing we have been looking into. And maybe from a research point of view, it all depends on the individual faculty member.

It's more the background environment. Probably MIT's situation must be the same. It's pretty much bottom up. Each individuals faculty member's interest is-- and everybody is sensitive, so they are introducing some new idea into the research. But we do not have any really strategic thing, except that we say that, OK, we emphasize bioengineering. Mechanical engineering saying that we would emphasize bioengineering manufacturing for a long period.

But bioengineering is certainly not just mechanical engineering subject. So the college started the department of bioengineering a few years ago. So it's the single independent department. So that that's a kind of new thing happening.

PRESENTER: More questions? Let's thank our speaker--


Our next speaker is Professor Dan Hirleman. He's a professor and department head at Purdue University. And he'll be telling us about the mechanical engineering and information technology at Purdue. Oh, and while he's switching over, let me just make a comment, which is Professor Tomizuka's talk reminded me that, after all, information has been in technology for a long time.

When I'm teaching the big undergraduate control theory course right now. And every control theory book points out to you that the whole reason that the watt is called the watt is not because Watt invented the steam engine, but because he invented the governor for the steam engine. The governor is a simple feedback device. Its goal is simply to process information.

And it was the governor that made the machine actually a useful tool, upping its efficiency, power, capacity by many factors of 10-- I'm sorry, by factors of 10 to 100. So I mean, information technology has always been in mechanical engineering. And mechatronics, I think particularly these beautiful examples that we just heard, shows you how that has blossomed as the capacity for information processing have grown. Professor--

HIRLEMAN: Thank you, sir. Well as he said, my name's Dan Hirleman. I'm at Purdue University. And I accepted the invitation to speak here and then saw the program and saw that I was last on the program. So I began to check into that and I know the reason now-- I have no connection with MIT.


Never been here. Actually, that's not totally true. I did an interview here, I guess you'd say, when I was looking to where to go to school for a PhD with John Hinkley when I was in the combustion business. But ended up choosing to go to Purdue, so maybe that's another reason--


The school of ME at Purdue has a long history. And just to put it in context with the MIT group, we graduate a little under 300 students a year. So we sort of graduate the size of MIT's undergraduate program each year. So we're three or four times that in size. We have about 50 faculty. Something like 300 grad students.

And we've definitely begun to focus on IT and how that will play into our school. I think that's still in process. One of the exciting things, opportunities we have, is a professorship, the Feddersen Professorship, focused exactly on that interface between ME and IT. So that's something we're in the process of filling.

Now information technology, don't want to exactly define it, but we sort of view it as where computing and communications come together, and ask the question, how does that influence mechanical engineering, both practice and education? Now if you look at the spectrum, another way of sort of thinking of information, this is kind of a progression. You start with data and assemble it in ways to make information.

In principle, you package information and we pass on or help our students acquire knowledge. I guess at the top of this food chain is wisdom. So information is down here in information technology. Another way to think of it is how do we package it and convert it into knowledge of our students?

Just a thought as we think about education-- this is something attributed to Benjamin Franklin, but out of a reference listed there. "Students are not receptacles. Information transfer alone is not education. Education is what remains after the information and training have been forgotten." So we do need to convey information, but we but we certainly don't want to stop there.

Now in terms of my talk, I think of sort of these four facets of ME and IT. And certainly there are many ways in which our profession contributes to the information technology revolution. We've heard some good examples of that just previously. The other side of that equation is that information technology certainly will impact mechanical engineering practice. And then there's a couple other educational contributions as well.

Now as the department head, when I knew that I was going to give this talk, the first thing I do-- it sounds like the same thing Professor Tomizuka did-- is you send out an email or chat with folks and say, I'm going to talk about IT and ME. Do you have anything you'd like me to show? And then I get about 500 slides.

So I'm going to move into that. I haven't checked this with chairperson, but I'm going to take a chance that you'll stipulate that I could show you a lot of slides of ME faculty that do really need things related to ME and IT. And I'm going to skip those. And if you don't believe me that I could do that, then look on the web afterwards and check it out.

But certainly there's a lot of contributions I believe our faculty have made, are making, and have made. Computation, in many ways, virtual sensors, heat transfer for the industry, data storage technology, wafer inspection handling, and the like. There's certainly a large contribution that our school and our faculty have made, as have the MIT and Berkeley faculty as well.

Now in terms of how we are implementing information technology in our education, this is a slide that I showed yesterday. I'm going to briefly talk about these, and then really kind of switch gears toward the end of my presentation. But we're trying to do web-enhanced, web-based courses. We've done distance design teams-- again, the communications aspect of technology enables that.

We have an interesting project teaming business and engineering students around some information technology. We certainly are including advanced simulation tools in our undergraduate courses. That allows you to do some more interesting things. And then we're certainly in the domain of doing virtual lab experiments.

Now in terms of web-enhanced courses, we have a number of experiments going, as do you. And most of you, I'm sure-- so we use streaming video and PowerPoint notes and chat rooms and frequently asked questions. And we think that is a way to certainly leverage faculty time and expertise and make a better educational experience for some of our students.

Mentioned the other day about how we use communications technology and some other types of information technology for distance design teams across schools, across countries, across companies. This was a project I mentioned the other day, yesterday in the panel-- a joint where we tried to team business MBA students with PhD students and accelerate that commercialization step into the last couple of years of PhD project. And we're doing some virtual lab experiments as well, as are a number of you.

I'll skip over these. Now actually for the rest of the talk, I'm going to take a little different approach. I think I won't come up with conclusions that are startling based on our conversation. I guess you'd say it's a different way of looking at the rationale. If you think about the impact on education and then delivery, I think delivery enables mass customization. The information technology and the potential in delivery can allow-- this is something that was talked about yesterday as well.

We may be able to tune our delivery to learning styles and that's effectively mass customization. I think it also reduces some of the repetitive aspects of faculty student interaction. These frequently answered question postings that we have have been a very interesting concept, chat rooms and the like.

So there are some things where faculty time can be diverted, I guess you'd say, to other types of things. And some of the aspects of faculty student interaction can be taken over by IT. It increases the reach of mediated material. You know, our school has I think 18,000 alums. If any of them do lifelong learning, I'd like them to take it from Purdue, not anywhere else. IT enables distance product realization teams. It enables this more realistic performance models in design and process.

I'd liked to now think about what I call some dislocation cases in businesses. So I'm sort of changing gears here and I want to step back and think about the impact of IT on some business sectors. And then toward the end, I'll try to bring it into our domain of engineering education-- in particular, residential university engineering education.

And these are some case studies I'd to talk with you a bit about. Encyclopedia Britannica, your local retailer, car dealerships, insurance agents, and then ask the question about the residential university. So let's look at Encyclopedia Britannica. Now anyone that grew up in the states-- I'm assuming that if you grew up in the US, you had a set of encyclopedias on your bookshelf at your house.

And my guess is, no one here didn't have that. How many one here has bought a set of encyclopedias in, say, the last 10 years? I don't believe anyone will raise their hand. So that's a very interesting case study. Very big business. Every American home, I believe, that had kids where their parents envisioned that they would go to college had a set of encyclopedias. And that's gone from 100% market penetration to 0. And it all happened just a few years back.

So what was going on there? Well, one thing is that the information conveyed in the product or the service/product was tied up in the physical books. So effectively, we all, all American kids, had about a meter of encyclopedias in a bookshelf at home. They used a parental guilt trip as a very effective marketing tool. Do you want your kid to be behind his or her peers? No. Then buy these encyclopedias. We all buy-- my parents bought--

Now what began to happen is not too many years back, Bill Gates went to-- my guess is, Bill Gates went to Encyclopedia Britannica and said, why don't we work together? And they said no thanks. And he said fine, I'll just give away encyclopedias then on CD-ROMs. And that's effectively what happened. Sales went to 0 overnight.

So interesting case study. Let's look at another one. Your local retailer. I think I can tie this to IT but let's think about it. Basically this is the small mom and pop shop. They have a relatively small selection, relatively small amount of things in stock. The proprietor carries the information content on these products. You went in, you chat with them, they know what's available, what they have.

They know a lot about their products. You know, you think of the hardware store. And there is certainly some social value, some community in that local shop. But what's happened, as you know, is the larger stores in the malls have begun to impact that. And you see they're not totally disappeared, but they're certainly an impact on that business sector.

Now one of the things that you might think about is, why is that? Well, one way to think of it is, as the malls are there and the Walmarts, basically the information available to the consumer has increased greatly. And so we don't really rely on that proprietor to give us information about the products. So that's one aspect probably of what's going on.

Think if you were in the car dealership business. And think of the trends. You know, many years ago, if you think of what's happened in the last 20 years, I think you see a trend toward greater selection with sort of fewer stops. So you think of auto malls where all the dealers that are competitors are clustering together in one place. I think certainly in the cities I've lived in, you find single dealers that are in-- now they have not only Fords, but they have GMs and international cars as well. So single dealers with a wider and wider selection of products.

Now there is this some people enjoy making the deal and enjoy the face to face battle with the car dealer and the information that they tell you about. There's probably some value in physically test driving these cars, but my guess is I would not want to be in the car dealership business depending on walk-in sales anymore. I think the internet opens up so much information that we don't really rely on the car dealership for any of that. We know exactly what they pay for the cars and we know all the cars that you can get and when you can get them.

The very last little case study I'll talk about is sort of your local insurance agent. There's a major dislocation going on in this business as well. You might, if you think back 20 or 30 years or something, I guess you'd say your friendly Allstate agent, your local agent, was the one that helped you if you had a loss of a car or a flood or tornado or hail or something like that. They came out. They helped you. And there was that sort of information.

Let me tell you how we can sort this out. And I think in time, you've gradually seen a bifurcation in that. I don't know. You know, the last time I got hit in my car, I drove it to a claim station who looked at it. My agent was never involved. Now I didn't like the way the whole thing was handled because it turns out I ran into somebody that had insurance from the same company.

And instead of arguing about it, they just blamed us both. And I didn't think that was quite right, so I called my agent. And basically my agent said, well, I don't have anything to do with that. You're talking to the wrong group. I'm the front end. You need to talk to some other division. So what's happened is that the whole industry has been broken apart. The claims is one place. If you want to argue about how they handled your claim, you don't call your agent. You call somebody else.

And I think if you read the paper, you see major changes that are going on in these large companies. And I guess the value added is in question. What does that local agent provide to you? So again, I think I read something that Allstate was doing something with a huge fraction and effectively laying off a huge fraction of their local agents.

So what are the themes in these case studies? And I hope they relate to what we're talking about today. Well, all of them had-- in the beginning, there were physical aspects of the product or the service. And the information content was integrally tied with the physical content. Think about the encyclopedias. All the information was in the physical books. Same with the car dealership. The information that the dealer or the agent conveys to you is tied with the fact that you go to the shop and see the cars or go to the hardware store and see the material.

Now the dislocation occurs when you can split the information content from the physical content of the product. I would say anyone that's working in a sector where your physical content and your information content can be split, I would beware. Now what's another sector that has exactly that potential? That's us. Think of a residential university. Well, our value added is the education and learning that our students extract from the experience on campus.

We facilitate it by faculty, and historically sage on the stage, good lectures, large classrooms, large physical plant, lecture halls, labs, student unions, dorms, study areas, all of that physical infrastructure that supports the learning environment that we think is important. But here's the question-- can the students, can our society, separate the information content of what we deliver from that physical plant?

And if so, we're in trouble. So a sobering thought maybe. But not a new one for you. So what do we do about that? So how does a residential university like Purdue and like the schools that are represented here deal with this? I think we-- and it's what we're working on, is we are working very hard to define exactly which educational experiences are really value adding. And I'm looking for the ones where the physical and the information content cannot be separated.

And I believe we need to focus on those very quickly because the University of Phoenix and other folks are going to come after that sector of our business where you can separate the physical from the information content. So where is physical presence not equal to wireless presence? Where is presence not the same as TelePresence? Those are the things that I believe. And at Purdue, we're trying to focus on.

So those activities. Hands on labs I believe are one of those. Now you might argue that I can do with haptic interfaces, you might argue that I can do virtual labs and do everything that a hands on lab does. I think in my heart of hearts, I don't believe that. Now it may be that I'm putting my head in the sand. But that's an interesting discussion. Will we ever be able to do with virtual labs what you learn-- when I talked yesterday about taking apart a lawnmower engine and putting it back together, every piece by piece and starting it and getting it to run, or taking part of a washing machine.

Can we really replicate that with virtual labs? Good question. I think a lot of-- I think mentoring, I think the faculty to student mentoring, that relationship, is one thing that you cannot do in a distance mode the same as you can do with physical presence. That's probably debatable. Some folks are going to be telling the world that's not true. They're going to tell the world, you don't need to go. You don't need to send your child to MIT or to Purdue.

They can stay at home or at least kick them out of the house, move them to an apartment nearby. But they can get everything they need wireless. So I think that's not true. Collaborative learning. I believe there's a lot of a learning environment that involves working with other students. That's the dorm. That's the midnight discussions. That, I believe, is a part of our physical plant and the physical and learning environment that cannot really be separated from the information content.

So when I think about building a new wing on the ME building, which we're trying to do, I think about how can I make that place, the place-- the one, the best place in the country and the world where this full educational environment can be experienced? I want the students to be fighting their parents to say, send me to Purdue, because I think it's going to get more and more competitive as time goes on.

Now that might be a little more provocative, but I guess the question is, what do we do with those aspects of our-- we're not a business, but those aspects of our enterprise that are subject to dislocation? One approach is say we just give those up. We're going to focus all our energy on those few activities where we truly provide a value that can't be split between the physical and the information content.

So I'm not proposing this, but that might say, let's get out of the internet web course delivery development, turn that over to the folks that are going to go after that anyway and let's spend all our time on these things. You know, I want to be the supplier of choice for the labs when all those distance education things realize that we can't get it across without a lab, then I think they'll be sending them to MIT and Purdue to get that component of their experience.

So I'm not exactly-- we're not actually doing that. We're doing a lot to develop web courses. But that would be one approach that a business would take. If you see a sector that's going to be eaten up, you just spin that off and let them fight on that battleground by themselves and we'll focus on our core value added. So--



HIRLEMAN: Be willing to answer questions, or maybe that's a piece of the discussion for the panel.

PRESENTER: No, as I said, the panel will be on applied issues, so let's have questions now.


AUDIENCE: Dan, I think maybe to your last point, you might want to consider what businesses consider-- something called asset management. And if you did that, maybe you'd see that-- how to leverage all of these things as opposed to disposing some and keeping others. Because really what you're really talking about is what are these assets and how to best use them?

HIRLEMAN: And surely the educational content is going to come from Purdue faculty and MIT faculty. So--

AUDIENCE: Well, there might be other business, academic interaction which could allow for that leverage.

HIRLEMAN: Right. Like we chatted last night.

AUDIENCE: I have a question about all of the virtual laboratories. And have you done any quantitative assessment of your success with them all?

HIRLEMAN: We have not in a scientific way. We have anecdotal evidence.

AUDIENCE: [INAUDIBLE] for an engine, we've had to add streaming audio in order to give people enough feedback. So then have your information about metrics or virtual laboratories.

HIRLEMAN: And my feeling would be you cannot do it. So my anecdotes go along with that, though it's certainly not an unbiased representative sample. And I also know that I probably bias a sample myself and then who you hear from-- so I think that's an important thing. I mean, to decide what the metrics are. And I thought the work we heard yesterday from Dave Wallace, I guess it was-- It was not on virtual labs, but the idea of how do you assess the effectiveness is a very important thing.

AUDIENCE: Yeah. Some of the comments you made and some of the questions were a lot of very good points. The one thing that is not there that separates whether you use [INAUDIBLE] MIT from the scenario to presented is that part of what we do [INAUDIBLE] well, we research universities, is we really research and graduate education.

And today, graduate education really depends on interaction amongst the people-- between professors and students, between students and between professors and between all of these people. And so since much of what we do-- in fact, the fact we're emphasising not only education so much in our department is somewhat taken to extreme in some ways.

But the fact of the matter is that we've assembled people together so that indeed through the interaction, we create things that really cannot be conveyed over the wire because it has not been codified. In fact, by the time we are ready to send it over the wire, it may be old knowledge. So graduate education would not be up to date. So that's a very important reason why schools like yours and MITs survive and do well.

But your point about letting someone else worry about transmitting existing information is a very good point. That process, this is something we need to think about.

AUDIENCE: If you followed some of what you've talked about now to some scenario-- one, I was wondering if your remarks on this idea of pulling together people from-- that are the leaders in certain fields together in a distance learning way to form a degree. So in other words, you pull the best from this institution and that institution. And that becomes the university. Is that out there?

HIRLEMAN: Well, before I move back to Purdue, I was at Arizona State. And there's an organization called the Western Governors University. And so that, in fact, is trying to do exactly what you say. And I can tell you, the legislators believe that the scenario you laid out can in fact happen and are moving in that direction and I guess you'd say competing with the universities for the funds that would allow you to do that.

So I think-- you know, and when I said spin off, that was overstated, but that's another way to look at what you're thinking. It may be that the top schools want to get together on that piece of our enterprise and do something proactive and get ahead of the curve. And so I've thought about that. I guess I haven't started-- I mean, I haven't started that conversation with anyone, but I'm interested in the conversation because I think that's something we may want to do.

AUDIENCE: Every time I hear about the top universities getting together, I'm reminded a bit of the European Union, where you know, there's all these countries and each one has a great history and so on. And they want to unite, but they still want to maintain their individuality and their culture and so on. So do you have some comments about that?

HIRLEMAN: Yeah, I'm not-- I don't believe it would be an easy process. And you know, it may be a different organizing group in subsets of the various ones. I'm not sure I'm recommending that or pushing that, but I think it is something to think about. We have, as I mentioned, all of us have alums who are loyal. And so certainly I'm not excited about giving up my alumni base for continuing learning to any of our-- as much as I respect Berkeley, that's not something I'm excited about either.

So it will be like the EU sort of thing, but it would be a long process. But I think what we need to make sure is we don't do nothing and then find out that we're in big trouble and can't reverse it.

PRESENTER: One more question.

AUDIENCE: It occurred to me that the context that you used to describe the issue is basically a market share problem. A pure declining market share. And it occurred to me also that maybe a way of looking at it to try to expand the served market. And it seems to me that the footprint of impact of the research university could extend on both ends into the high schools and into the early career. Then that might be a way of looking at the student life cycle differently. And it might be a solution. I have a need for a solution.

HIRLEMAN: And I think all of us are working with the high schools, pushing everything down that direction. And I think we're all thinking that. So that's a good point. That's true. I was kind of painting a scenario where you'd say the residential bodies are going to decrease. That would be the worst case scenario. But I don't think that's probably going to happen. I think the students that go to the major universities and live there are probably still going to.

PRESENTER: One more. One more question.

AUDIENCE: I really think the fundamental issue is going to be the residential university system is extremely extensive. Parent [INAUDIBLE] educate their children. And the apparent pre-cost of the price you pay for, quote, distance type learning is such a huge gap between the two, that I maintain that will ultimately be the fundamental driving force which will make residential universities have to stand up and figure out a way they can offer something in a far more cost-effective place.

HIRLEMAN: Or convince the world that the value in-- or focus on those aspects that clearly add value to a residential university.

AUDIENCE: Every one of your examples was the rule way not swayed by the [INAUDIBLE] personal contact with a local community.

HIRLEMAN: Well, Purdue's ME enrollment has been flat for 20 years. And it's capped for 20 years. So I guess I haven't seen the arguments lose. But I am trying to be prepared for the fact that it's going to get tougher.

AUDIENCE: At MIT, we have a large number of departments without any undergraduate students.


PRESENTER: There's plenty of it. All right, let's thank Professor Hirleman again. Last on our agenda is our panel discussion. It's listed on the program as the role of mechanical engineer in the information technology [INAUDIBLE]. And I would like us to interpret this broadly just to talk about all sorts of applications of information technology and mechanical engineering-- information technology and mechanical engineering and vise versa.

And you know, now's your chance to speak up. We should get a discussion going. Our panelists are going to be [INAUDIBLE] from Ford and then we have our speakers from this morning. Professor Tomizuka, Professor Hunter and Professor Barbastathis. So if I could ask you gentlemen to ascend the stage here.

Now [INAUDIBLE] is the only person who hasn't spoken yet. Actually, is he here? Well, if he's not here, then he won't speak. That's strange.

AUDIENCE: So he just stepped out.

PRESENTER: It's possible he just stepped out. Yeah. Well, if he's not-- why don't we get ourselves going then? And so if people want to say a few words to some of their feelings about things-- or not, as you please. And then we can just have a discussion.

AUDIENCE: We have some questions.

AUDIENCE: I'm going to start with an old, embattled question. Should we be requiring a programming language of our undergraduates? And if so, what?

PRESENTER: Whoa. There's a good one.

PRESENTER: And then where's Sunny Siu? That's a question-- he's on-- yeah, do you want to address that, Sunny? You are the university--

PRESENTER: [INAUDIBLE] asking that question directly focused [INAUDIBLE]

SIU: I'm teaching undergraduate courses-- sophomore, freshman level internet courses. Last year, there were about 20 mechanical engineering students. This year, there are like, over 40 students. And it is not a required course in our department. It's an elective. And they all want to learn about this technology.

And the student evaluations, evaluations from students I got, they always say that is something that they want to really see in the curriculum because they feel like they are really the top notch in the forefront of technology and they want to see how this is going to be linked to mechanical engineering. So one thing that-- of course, if the course is elementary, I cannot go into all the specific application, like telerobotics or telemedicine or this kind of application.

But I do [INAUDIBLE] at the fact that the internet is more than just a tool or a network to retrieve information. It's about to control machines, to connect with physical objects. And in that course, I did teach them about Java programming languages, XML. They'll be used in a lot of medical databases. So this kind of thing that we can teach the mechanical engineering students without giving a whole course about c++ or object oriented programming. So that's the answer to your question.

AUDIENCE: Professor Tomizuka, what did Berkeley do about--

TOMIZUKA: I think what's happening, we don't teach the programming language in the traditional sense, but I think we really are still requiring a course is essentially to think about something more like what solving-- what the computation is. And also I think I believe I'm not directly involved in, but I believe that we teach also c I think at a lower division level.

Another thing I think really if a student is interested in certain language like Java, it's-- one of my colleagues, he's giving a new course on Java programming application. And I think he was very surprised that he-- it is the first time he's giving. And it easily went over enrollment limit. So instead of-- from a faculty point of view, whether we should be teaching or not, certainly I think students are interested in learning some language aspect. And I think probably to some extent, listen to customer, what they want. And I think, again, for ME department, somebody teaching Java language and getting lots of student enrollment, it's a very, very good thing for us.

PRESENTER: At MIT, it's civil engineering, right? Civil engineering is by far the most popular courses there for 100, 1-0-0 C++. You and Connor? Yeah.

AUDIENCE: Yes. My name is Martin Brouillette from the University of Sherbrooke in Canada. I think one of the conclusions of this weekend's meeting is that information technology is going to be very important-- it's important now and it's going to be more and more important in the future. And we've been adding courses piecemeal into the curricula, but what we have not done yet is integrating that to the training of Mechanical Engineers, because in the past, we used to say that fluids and solids would be important.

And now we're saying that idea is important, but have not seen it integrated as these past subjects have been. What's the future of that? Maybe you have something to say about that.

PRESENTER: I'll make a few comments related to teaching instrumentation where we have the challenge of integrating optics with mechanics, with electronics, with chemistry, and also normally biology. One of the problems is certainly in those courses, we expect students to have a basic numeric language such as Java or Basic or c++.

But we also find it highly necessary that they have some symbolic capability. But it goes way beyond that because-- I'll just give you an example of a challenge that involves multiple disciplines. Imagine you have a conducting polymer that is clear. The polymer is clear, but it's also an actuator. So it's electrochemically activated to contract and it's acting as a lens.

So there you've got an issue of needing to model, normally with a continuum model, the electrical activation of that polymer, the chemistry involved with it, the fact that there's mechanics causing this material to contract, which has an optical consequence, namely that it's a lens. And because the efficiency is not 100%, you've got a thermal management problem as well.

So the problem is that the existing modeling tools, be they lump parameter or continuum tools-- think of your standard finite element packages. You'll have one package that will be appropriate for mechanics and thermal, but normally there'll be no optics and no electronic phenomena. Or then you'll go to some electronic finite element or continuum modeling package, and it'll perhaps include magnetic phenomena, but not optics.

You go to optics and there's no electronics. So one of the problems at the moment is that as we go to more and more sophisticated systems involving information manipulation as well as energy manipulation and transformation, the existing modeling tools are rather fractionated and are not well tied together. And I see that as a tremendous challenge to put together very generalized modeling tools where simulations and virtual environments and so on can be created which go across these different material domains.

AUDIENCE: Actually, if I could make one more comment-- there's one more area where similar challenges exist. And that's MEMS. MEMS is micro electromechanical systems and has kind of been taken over by electrical engineering departments, if you look around. Yet MEMS is a very good example of integrating requirements from electrical, mechanical, and even optical properties, optical MEMS.

So again, you have the same challenges that [INAUDIBLE] was mentioning. And then the question is, what do we want our students to do? Do we want them to just use naturally computers as other disciplines do-- doctors, lawyers, I mean we communicate with them with e-mails nowadays, right? Is this what we only want them to do, or do we want them to go beyond and contribute to modelling and contribute to-- or even more fundamentally, new advances? Harry had a question.

AUDIENCE: Yeah, I just [INAUDIBLE] on questions regarding how IT is still being mentioned, whereas the mechanic engineering-- because it's very critical questions. And just going back to Professor Hunter's comment, this has been a really important issue for us in the [INAUDIBLE]. I think, as [INAUDIBLE] mentioned, that the data are moving to information and then transforms the knowledge and finally wisdom.

And that does provide a very good framework because we at mechanical engineering focusing mostly on the knowledge and the wisdom part. And our mainstream discipline has been both in that way. But it's a time that we need to have some new format, representation, of these stuffs and the way that pieces of input, knowledge, to be used.

For instance, time is a very critical factor. We have to make very critical design decisions overnight. E-commerce is a major driving factor in doing that. And then that's actually giving us the big challenge as to we can do slow dwelling onto something. But we had to use these things very effectively. And I see that the traditional mechanical engineering hasn't been prepared for this changing.

As far the presentation of the format is concerned, I don't think that's it's very well prepared for it. I think Professor Hunter has developed a very interesting course. It's called-- do you have change in course name? But 2.131.

HUNTER: It's called Advanced Instrumentation and Measurement.

AUDIENCE: That is heavily based on simulation and modeling and connected to a real data acquisition, which is very important part. And that's one form of a very important style of mechanical engineering. And I think we haven't done materialized that in the broader undergraduate subject yet. But I know students have to learn these tools. And then they may find some ways of using it for their future career development.

So I think the connection is very obvious, although we haven't done that. But talking about all equal parts and stuff, but we have to look at the issues beyond just the merchandise and the purchasing stops. But some important engineering level issues hasn't been explored yet. And that's a kind of a point that we make a strong connection between the IT and the mainstream mechanical engineering.

PRESENTER: More questions? Points you want to raise.

AUDIENCE: I wanted to come back to your comment again about the software, that a lot of the tools aren't there. It's an intriguing thing that I thought because right now we have the ability to integrate solid modeling through finite element analysis with some of these tools. But historically, we've resisted that because there was a feeling that that's a black box and we want our students to understand the fundamentals.

But it's clearly a reality, because that's what industry is using and we have to integrate it into the curriculum. If your point about how the research there is then leading us to be able to get much more in a given tool, which combines all of the, quote, a lot of the fundamentals beyond what I just said earlier, is rather intriguing in its effect on education and what we talked about earlier about distance learning or anyways, very early intriguing what that will do. It's just a comment, trivially, [INAUDIBLE].

HUNTER: I think that we have some enormous challenges ahead because there's a sense in which you could say that the easy pickings are over and a lot of the really interesting research and commercial systems in the future are systems. And they're going to be drawing upon tools and techniques from optics, which has classically been in physics, through electronics and mechanics and new materials and information processing, and so on.

And so the problem is, how do we train the sort of Renaissance style systems level designer and scientists and engineers who can handle these sorts of systems? And I think the problem is unless we can increase the efficiency with which learning occurs or with which information is acquired, we've got a terrible time allocation problem.

And I think we've got to make some hard judgments as to which areas of knowledge are perhaps to be left out in order that we progress forward. I'll give an example. All of you, either in your calculators or in your favorite computer languages, will use the basic trigonometric functions, sine, cos, tan, and so on. How many of you know algorithmically how that particular function is coded in your favorite calculator or in your favorite computer language? Could you put your hand up?

AUDIENCE: I only know because you told me.


HUNTER: Then you know that the favorite-- and for some of you who didn't see it, probably about 10% of individuals put their hands up. For those of you who don't know, it's normally a rational Chebyshev expansion where the objective function is not only squares. It's a min max criteria. Now very, very few students know that now, whereas 20 years ago, you were taught that when you were learning about writing compilers for typical languages.

You would know that. And the same is said for the Fourier transform. It used to be that you showed your metal by having coded a Fourier transform from scratch. And then having done it once, you were then free to use, to plunder, some package for future use of the Fourier transform. Nowadays we have to ask ourselves, do we have time to show students the elegance of setting up your butterflies-- in the Fourier transform butterflies in a certain way, or do we have the time to expose students to the fine detail in a singular value decomposition algorithm?

And if we do spend time on singular value decomposition and all the details on how that magnificent algorithm is implemented, we lose the ability to teach some other concepts. I think we've got a challenge in terms of making these trade-offs. But I'd hope that at the same time, we don't assume that the efficiency is constant. I would hope that there are, via using new technologies, methods of increasing the efficiency and speed with which information is acquired.

AUDIENCE: How do you define efficiency in learning?

HUNTER: Well, that's a very difficult thing because at the moment, there's a sense in which if I went into my laboratory, used a multimeter, which, instead of having five significant places, I had none. In other words, I perhaps use my own hair as an electrostatic volt meter or something like that. You would ask me, what's the reliability of that measurement? What's the validity of that measurement? What are the signal to noise ratios?

In general, we don't ask those same questions of the process of transmitting information. We don't have the measurements. There's a sense in which in a scientific sense, it's a very primitive-- our measuring tools are very primitive. And Dave Wallace and others are attempting to improve those tools. But the problem is that we have great difficulty even quantifying what we mean by something like efficiency. So all I can do is fall back to anecdotal feelings about it.

But I do know that in many of the discussions that go on about how we allocate time to different subject areas, there's almost an implicit assumption that the efficiency with which information is acquired is constant. And I don't think that's the case. I think by-- it's very, very clear from the behavioral learning literature in animals, for example, that you can craft situations where the speed with which an animal will learn a simple task, the efficiency with which they acquire that information, can be orders of magnitude different if you set the environment up appropriately.

Whether or not humans are subject to the same principles, I don't know. But it just strikes me that we have to be looking for ways to speed up the rate with which information is acquired.

AUDIENCE: I guess part of the complication is you think of levels of learning, it's easier to assess the lower levels, which is the training example. But I think it's harder and harder as you move to knowledge and opportune levels of learning to assess.

HUNTER: Yeah. And in fact following up on your point before, we have the concept, for example, in the advanced instrumentation and measurement course that, as Nam was saying before, we're exposing the students to the culture of a research environment-- that by actually running a course not in a departmental laboratory, but in a real research laboratory where things are being fabricated, measurements are being made, things are not working all the time as they might in the departmental laboratory, you're exposed to this culture of research.

Now how do we quantify something like that? I think it's very difficult. But somehow, it's special and distinguishes us from what might be transmitted across the web for example. Sorry, interrupted you, Nam.

AUDIENCE: Just in response to Alan's question here-- I think we are really trying to measure how much people retain certain knowledge so they can use it. It's not simply a transmission of it. And if you look at it like that, as someone who has this much knowledge will be able to add so much given a certain amount of information. Someone else who knows this much will we able to learn more given the same amount of information.

So that means it's sort of like an exponential function. The only way you can measure that is coefficient of the exponential function. And yet we are trying to treat it as a linear thing, thinking that there is-- everybody has all acquired same increment. And so what we haven't figured out is what the e to ax is what a is. And a t or whatever the case might be, right?

And I think that's why they sort of want to deal with talking as measurement of the teaching efficiency or learning efficiency, because it's not a linear function. And then every student who has different levels of understanding or levels of knowledge base will acquire different amounts given the same amount of information. And that makes this whole subject very complicated.

PRESENTER: Actually, I will take a pretty radical position. That is that part of our job, of course, is to teach the students very fundamental things. And there's a certain number of things that they have to learn. But part of our job is also to teach them how to learn. And I think very often we make this mistake where we try to pack as much information as possible in our lectures, in our syllabi and so on.

And what we miss then is the students actually get swept by all this and they retain very little. On the other hand, I mean if I look back at my own background, the topics that I learned the most from were topics were the actual content was very sort of carefully picked, but not sort of universal. But they were very some topics that were very well done, very deeply done. And in addition, the instructor had the intelligence to arouse my curiosity about the topic.

So then in addition to whatever I was doing in the homeworks and so on and so forth, I would also go and look elsewhere. Because if you think about it, we don't have to transmit all the information to the students ourselves. Books have existed for hundreds of years. And that's where the information used to lie to exist. And now we have the web and all kinds of others. CD-ROMS and so on and so forth.

So on the other hand, if we manage to arouse the motivation or the curiosity of the students, I think we could achieve better efficiency. And I think this is how this can be done.

AUDIENCE: Can I say you ask Professor Tomizuka to give a comment-- you have this beautifully crafted rather personal mechantronics controls.

TOMIZUKA: Control. I think control is an interesting area to watch because if I look back my 25 or 30 years from student to now, I think there's a control area always evolved with information technology. And so when I came to the United States, in fact, I think I have already mastered quite a bit of control theory. But the first thing I experienced, I think taking a course 2.151 at MIT was [INAUDIBLE] there.

Now there was this homework assignment to check step response. Now that last transformation, inverse transformation, time response, and so forth, I thought I knew, but first time when I really got some good feeling was with [? 1130 ?] solve the homework problem. And actually, the computer I started plotting the step response. So I think there's a information technology or use of the computer, if it's used in the right way, it's really enhances how a student would be understanding.

We cannot really say that because the information technology is there, we cannot become a slave of the information technology. Everything is done by now computer, whether we don't have to teach on any particular aspect. That's a total mistake. The controls, we still teach some of the very old graphical technique which was invented when the information technology was not there.

But by some of those techniques, really helps us develop good intuition into this system, engineering systems. Some of even the people called modern control theory. There are lots of matrix involved. Just when I'm watching matrix, I can follow the theory and I can come up to that right. But I don't get good feeling when I get some numerical work done and see how that past certain parameter affects the system performance.

That's when I really get the theory. Really I am fully and I understand and make it a sort of working knowledge. So it's really an important aspect to really combine the fundamental thing and utilize computer to really increase the understanding.

AUDIENCE: I was wondering if any of you could comment of this idea of virtual laboratory in the following the way-- we talked about one of the important things at university is the hands on, the mentoring, the interactions. Now the virtual laboratory has the potential to replace somehow that. Does anybody have any evidence to suggest that that's as effective as a hands on laboratory.

HUNTER: Well, I don't personally have any evidence to the extent that we've operationally defined what the variables are and we have good signal to noise ratio and the measurement techniques for acquisition of information via direct interacting with interaction with phenomena versus that you might get across in some virtual environment.

However, what we can say are there are certain phenomena that, at the moment, the technology does not allow for transmission. For example, when you take the conducting polymer example that I gave before up to 10 to the 7 amps per square meter, the smell that occurs as it burns is not something that currently is convenient to transmit across the web. So we have good tactile interfaces now. We have very good 3D visual interfaces.

But for example, the chemical display technology is relatively primitive. So when you go across different sensory modalities that are involved when you really got your hands in on something, you're detecting the moisture content and materials. You're feeling the slipperiness of bearings and so on and so forth. You're smelling the consequence of bearings being burnt out or conducting polymers being overloaded.

These are things which, with time, there will be presumably good ways of being able to transmit that via a remote interface. But not at the moment. So I actually don't think that we should make a clear dichotomy between the two. I think in many teaching scenarios, you want both running at the same time.

So for example, we and others in our department are conceiving of courses where you have wearable heads up displays and wearable computers so your hands are free to interact with machine tools and other tools and where the display technology is in one instance see through so that you can see what you're doing, or when you need additional instruction that information comes up in that same display.

So I see mixtures where real laboratory work is being undertaken at the same time as portions of it is done virtually.

AUDIENCE: So we need to develop web-based scratch and sniff technology.


AUDIENCE: They [INAUDIBLE]. No, but I rarely mention this because I always get a lot of guff for it, but I actually was an undergraduate at Harvard and-- you're not supposed to say that at MIT.



AUDIENCE: At a university in Boston. And there they let you-- in the physics department, you could either take a lab course or they would let you work on an experiment instead of working on doing a lab course. And I went and I worked for Norman Ramsey in France on his cold neutron experiments, for part of the reasons that he actually won the Nobel Prize three years ago.

And so I never took a laboratory course as an undergraduate. What I had to teach the first semester I was here under Ian, I taught 2.671 lab, which was a laboratory. And it was such a different experience from what a real experiment is like. I mean, in many respects, it was better, because of course the techniques are being taught in a systematic way. But I felt that it really-- this is not even a virtual experiment.

This is an actual laboratory where you are putting stress gauges on things and stuff like that. I felt much of the experience of actually being out there during research was really lost because you're being presented the stuff, if you do everything right, it happens in the way it's supposed to. This is a problem even worse with virtual reality stuff. And most of the experience of an experiment is sitting there frustrated, typically for months at a time, when you can't figure out what's going on.

Trying everything, trying to think of all the things you didn't think of in order to make things work. And so I think that if we actually want to educate our students so that they are these kind of Renaissance people that Ian was suggesting, I think that going to the point where they're in a more controlled and less sensory rich environment is going in exactly the wrong direction in terms of teaching them the skills of imagination and dealing with the real world that they're going to need. They're going to actually do something good later on. And that's just my prejudice.

AUDIENCE: I had a few new comments on the panel from [INAUDIBLE] about the impact or implications of IT on one traditional aspect of mechanical engineering, which is design and manufacturing. We know that the IT facilitate the universal and instantaneous access and dissemination of information-- in this case, the design and processing information. And that give a new life of discipline like concurrent engineering or collaborative engineering. So I'm wondering if MIT, UC Berkeley, Purdue, or other institutions have incorporated the IT into that kind of a discipline?

AUDIENCE: Perhaps Sanjay--

AUDIENCE: Yes, Sanjay. Excuse me.

AUDIENCE: [INAUDIBLE] The design and manufacturing and educating in the IT realm. Berkeley, of course, is doing a lot of work. Obviously, [INAUDIBLE] design [INAUDIBLE] go and design [INAUDIBLE] university. In the IT, there's the [? Door ?] project, which is [INAUDIBLE] current design where you can connect to catalogues online and run models, for example. If you have a model and you want to plug it into your design and see if it works, I can actually download it, the model, [INAUDIBLE] talking about something for the model [INAUDIBLE] a way to simulate, cross-simulate systems from a distance and also contributes to concurrent design.

In my research, we're working on haptic systems. We built a the first fully three dimensional haptic system actually to do two-part generation and to design objects in a CAD system. So that's really incumbent on the fringe of IT. So I think that at MIT, at Berkeley, at Stanford, there's the CDR-- Center for Design Research-- has done a lot of work in agents, using internet agents, to do concurrent design.

So there is actually a lot of work of IT. In fact, I would say that's the first area where IT made an impact is in design and manufacturing, because it is the most convenient way to disperse information. But I also wanted to make another comment about the question that was asked earlier about experiments and can we get rid of experiments and have virtual experiments.

As Ian pointed out, we've had some trouble-- you know, you can't completely remove that visceral experience of being present at an experiment. So at MIT, we're trying to find ways to do the lecture delivery by the internet, but the experiments more physically in the lab. But one thing that Professor Mary Boyce did was portable experiments, which is take home experiments, so that you can put together small kits.

For example, tension test specimens that you can break at home and measure at home. Similarly, there are kits that are put together for a 2670 project in-- I'm sorry, the six-- in electrical engineering, there's a project where they can actually build electronics at home. So that may be another way to make experiments portable and maybe address the same question.

AUDIENCE: I think we had a question that the [INAUDIBLE] university, about how we defend ourselves in the face of IT revolutions. Certainly, we lose some of the undergraduate programs to be replaced by that, and how we find our niche market. It seems to me that it is really a real posed question to some extent, because the technology side is ever increasing. And as [INAUDIBLE] pointed out, the huge gap in the cost that can't beat.

On the other hand, I think the real important issue here is it's not the kind of stuff that our kids can do because of having a large amount of information, here's the-- on quick case is really a research part. And I confess I do have several of my students having learning-- not a learning disability, but a research disability. And even MIT students that will pass the qualifying exam having a research disability.

And many of them are smart kids and they know a lot. And then as a matter of fact, I see that over the past several years, they're collecting information very easier. And then they are spending every evening in front of the internet and then they learn a lot. But I know doing some research is not the kind of stuff. And sometimes I ask them, put all the stuff that aside. Just take a pad and pen. And just concentrate.

And those students sometimes they're having a sort of a difficulty to concentrate. And then of course, the future leader should be able to deal with the large amount of information, yet they must be able to concentrate on one thing. But in the course of educating these younger people, I sometimes see that the-- not the IT, but that DIT is better than exposing them to any piece of information. Sometimes I request them not to read, not to explore, not too learn, but information they have is just good enough.

Just look at yourself, trust yourself, and then think about the stuff that we already discussed. Sometime it works, sometime it doesn't work. But I know in my experience, that's really pointing the-- one of the important aspects because creating something and actually inventing something is really different. Of course, a broad knowledge is very essential. However, that does not-- that may be some necessary condition, but it's not sufficient. And it's really lacking something. And then when we plan the future of a university like MIT, I think we should have kind of good mechanisms in doing that to assess the first representation using--

Well, yesterday we saw a better version. You use the OHP. But usually he's actually a chalk talker and doesn't use any IT technology. But just sketching, it's just a few icon or a few creations, so to speak. But that transmits a lot of information. At least, a kind of basic concept is better described by clicking on the-- I shot about 100 [? Bluegrass ?] yesterday, but sometimes such mode is better than my mode.

So that's actually point out these important things. In the face of IT, what we research institution and educational institution should convey to to our students. And then that's really a sort of a joy of inventions or creating something better now on the knowledge. And I see that the kind of stuff that we share with the students is really very much a emotional component or kind of a sense or detection abilities that Ian pointed out, which are totally different direction from IT. That I'm looking.

AUDIENCE: I'd like to add to what MIT kind of was-- put him on the spot for a minute. How important is it in your work to have the MIT undergraduate program, and why?

HUNTER: I can answer that. First of all, I just love the procedure we have at MIT of having every undergrad do the same courses initially-- physics, math, chemistry, and effectively molecular biology. I think that's a great underpinning for almost any discipline. Secondly, the Undergraduate Research Opportunity Program, the UROP program that I think you're perhaps referring to, where students are exposed to real research laboratories from day one if they choose to-- and I believe, Nam correct me if I'm wrong, but about 90% of our undergrads do UROPs.

AUDIENCE: Yes. Somewhere around 70%, 80%.

HUNTER: Yeah. So in my laboratory, we have a very strange situation or strange for most universities where we have the same number of postdocs as we have PhD students as we have master's students as we have undergrads. And it turns out that some of those undergrads can be, at times, as effective as a master's or PhD student. Some of those kids, if you put them in the right environment, are incredibly creative and productive.

And so they're not just there because I feel good about having undergrads. They are, in fact, productive in a research sense.

AUDIENCE: Undergrads don't know that certain things--


HUNTER: Exactly.

PRESENTER: It's true. I think a good-- I would second that. A good UROP is worth a couple of PhD students at some times.

AUDIENCE: Yes, if they're motivated properly.

PRESENTER: If they're motivated properly. Correct. Well, let's let this be the last comment [INAUDIBLE].

AUDIENCE: With all of the emphasis going on now in K through 12 on having computers in every classroom and 100% access to the internet, what are the panel's thoughts on what it will be like five years from now in terms of the idea that you will have students coming in undergraduate that maybe have 10 years of IT experience?

PRESENTER: [INAUDIBLE]. Would you like to comment on that? Go ahead.

AUDIENCE: I mean, will that change the scope of the problem?

PRESENTER: It definitely will. If you wish, it's a matter of satisfying the needs of our customers, but also taking care that we don't bore them, so you know, if the kids come in and they already how to program or they already know how to use computers, why else would we bother doing that again? Unless there's particular gaps in their education the way they come in. So there's no doubt that universities will evolve just because of that.

Of course, still when the kids, they come in, what they don't know is they don't know how to be in an environment where everyone else is equally as smart as they are. They're coming from high school where they're probably the best in their class. So they come here and they have to learn how to deal with that. IT is not going to fix all these things. Even if you think about virtual interactions, and so on. But in terms of making our job easier in using IT, that's definitely true. What you're saying is definitely true. Ian, you want to add something?

PRESENTER: Why don't we thank our panelists. Thank you very much.


All right, so now I think I've scheduled closing remarks. And you're probably pretty tired of hearing me talk. So let me just make this brief to summarize what my perception of this meeting. So now let me let me do this by just talking about something I talked about briefly before. We have this thing called-- the phenomena that's really driving information technology revolution is this Moore's Law phenomenon where I insert the size, 10 to the minus 1 meters back in 1950.

You know, 10 to the minus 7 meters now. Now in 2000, in Y2K, and we don't know where we're going to be going in the future. If I draw the-- of course, this tells me the size of components of things, but if I look at the power of computer, naturally the power is proportional to the number of components you can put into place. So the power is going in the opposite direction, back from-- you know, 10 to the 3 that back in, say, 1950 where you know, now we're 10 to 12 or something like, well, something if we include our memory space, it would look like that right now.

The power of this is increasing rapidly. This is just a fact. Along with this has come a huge increase in our capabilities, the kinds of things that Professor Tomizuka was talking about in mechatronics, our ability-- I would not have not thought it was a better idea in a Xerox machine, you know, have the paper handed off more times. But if the Xerox machine is a smart Xerox machine, then maybe it is. And let's hope it is, because Xerox machines are about the most clunky things in the world.

All right. Now let me give-- I think that in some sense, this conference throws a very interesting light on Moore's Law. Now my first graduate degree was in history and philosophy of science. So indulge me for a second while I take a historical and philosophical perspective on this phenomenon. Our conference is about ME and IT. And we heard a lot of how mechanical engineering from--

We've heard a beautiful talk, an initial one, by [INAUDIBLE] about how mechanical engineering had a big effect on the information technology revolution. And I think that this conference has really thrown light on that. Now we've also we've heard even more-- in fact, considerably more-- about how the information technology revolution has had an effect on mechanical engineering.

Now you don't have to be teaching control theory to realize that this is a feedback loop, right? And so if these kinds of things that happen here is that if we have-- we have the rate of increase of improvement of information technology, VIT CT, is proportional to, is equal to A times ME. All right? And D ME DC is equal to B times IT-- all right, what happens in this case?

Well, let me see if I can-- I'm really bad about doing algebra in public, but I think I can do this. This says ME and IT as a function of time, they go-- let's see. E to the square root of AB times T. Both of them grow at this rate. You know, I just have to plug this equation into this other equation and I get the second order equation. And then I have to take the square root.

But all right, it's an exponential explosion. That, in my mind, is what gives us Moore's law. And Moore's law is not just about information technology. I think we've heard all this beautiful work from Ian Hunter, from Peter So, from George Barbastathis, from-- I could go through the list of all the people about how other kinds of technologies like instrumentation technologies. The precision of information technologies is also going down at an exponential rate.

Moore's law is not just a phenomenon about information technology. It's a phenomena about all technology. And what this conference has largely been about has been taking advantage of all these things. So we have this exponential explosion. This is what's going on. And I feel that the technical parts about this conference really illustrated this quite nicely. Not to pat us on the back or anything. I think it's just a fact about the world. It's not a fact about research at MIT. It's a fact about technology in general.

All right. Now then we had, for me, a very provocative discussions of education. I think Alex d'Arbeloff really brought up the question quite nicely-- do we take a software engineer and teach him about mechanical engineering if we want something done, or do we take a mechanical engineer and teach them about software?

Well, most of the observational evidence here is that you take out some really smart people who are great technologists and they're going to learn the software and make that happen. However, we also, I think, have a very good example-- heard a very good example of an alternative, which is you take some people who are really good at software and you take some people are really good at engineering and you have them collaborate together in a team.

A superb example of that is the auto ID center, where we have a world expert on the internet, Sunny Siu. We have world experts on creating nanotechnology for tagging purposes, like Joe Jacobson and with Sanjay Sarma all working together to create a fantastic enterprise, which would not have been conceivable a few years ago. Something where you could, in principle, tag every atom in the universe if you could find a place to stick the tag.

Unfortunately, it's hard to stick tags on atoms. But now then we had a very thoughtful set of discussions really primarily I think from all of you together about these issues of education. I've been very happy to-- I've learned a lot from this conference. Not only did I learn a lot about my colleagues work, since I never hear about except when I go to conferences. But I learned a great deal about your attitudes and thoughts about the educational process.

And I was very happy to see that there's a kind of a congruence here. I mean, we learned something I think that's very important-- that first of all, clearly as Neil was saying, education, distance education, the potential for information technology in education is vast. You could potentially have inexpensive education for lots and lots of people.

But then we got a great presentation from Dave Wallace, who you know in addition to just talking about this has done these actual experiments, which is kind of nice to see data, where he pointed out that actually information technology-- sorry, distance learning on its own via the web and just sitting there going to lectures is kind of comparable. Lectures are just the oldest form of information transmission. They started in the University of Bologna in about 1,000 years ago.

And a lecture then-- they didn't have printing presses, so the lecture was just the lecturer reading a book while everybody copied it down. Just word for word. Now it's changed not very much since then. That's a thousand year old technology essentially at work. Thank god we finally are getting something where we can enhance it. And Dave Wallace's work I think showed that.

If you take information technology and you combine it with teaching, put them together, you can make people do better. Unfortunately however, it cost 500 hours per hour of lecture. It's a lot. It's going to be very expensive. The investment is going to be worth it in the long run because of the economies of scale that Neil was talking about.

So I think that-- and then I think Dan Hirleman I think gave a great talk about this. I was amazed to find the kinds of-- the words that we've been hearing here at MIT talking about how we're going to combine information technology with engineering. So word for word coming out of someone from another university with a much larger mechanical engineering department, as I understand.

So I think that I'm willing to declare this conference a success. Why did we have it? Well, clearly we wanted to show that at MIT, we're very concerned about information technology. The department has made a lot of investment in terms of faculty to work on it. I don't think there's any question that information technology is going to play a huge role in mechanical engineering in the future, as in all engineering.

It's not really something to debate about. And so let me just finish up. And we're going to be there to see this. These are exciting times to live in. Let me just finish by thanking folks. Marion Gross, who's not here, is the secretary for this. I think we should give her a hand of applause.


Remember Sanjay Sarma, the Julius Caesar of our triumvirate who really did all the work for this.


Sunny Siu, my fellow warden at this event.


Nam Suh who has ordered us a fantastic clambake last night, but whose idea this was in the first place and who came to us and suggested that we do it. And we said, ah, do we really want to do this? But in the end, we decided it was a great idea. And I think that it would never have happened without you. So thank you, Nam.


And from my point of view, most important, all of you for coming and giving us input. We're all in this together, even if we're competing for students and influence. The world is changing because of the information technology revolution. People in this room are making it change.

AUDIENCE: And we have to thank you--