Science Reporter; 42; Computer Sketchpad

I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry I'm sorry, I'm sorry, I'm sorry I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry, I Whether this

bridge will be strong enough to withstand the loads it may encounter, but instead of first having to translate the problem into numbers, the only language on computer understands, this oscilloscope together with advanced computer programming, make it possible for the designer or engineer to communicate directly with the computer in graphical terms, the language the man best understands. This computer sketchpad is our story today on Science Report water. Hello, I'm John Fitch,

MIT Science Reporter. We're at MIT's Lincoln Laboratory in Lexington, Massachusetts, a major research organization which has played an important role in developing this nation's defense system since World War II. And although Lincoln Laboratory is perhaps best known for its work in radar and advanced warning systems, it has also seen the beginning of the computer era and has made many important contributions to it. To learn about some of its recent work in improving the relationship between man and this important machine, we talked with Professor Stephen Coons, an associate professor of mechanical engineering at MIT and co -director of the computer -aided design project. John, we're going to show you a man actually talking to a computer in a way far different than it's ever been possible to do before. Surely not with his voice. Now he's going to be talking graphically. He's going to be drawing and the computer is going to understand his drawings. And the man will be using a language, a graphical language that we

call sketchpad that started with Ivan Sutherland some years ago when he was busy working on his doctoral degree. And you will see a designer effectively solving a problem step by step and he will not at the outset know precisely what his problem is nor will he know exactly how to solve it. But little by little he will begin to investigate ideas and the computer and he will be in cooperation and the fullest cooperation in this work. Now how does this differ from the way the computer has been used in the past to solve problems? Well the conventional way, the old way of solving problems with the computer has been to understand the problem very very well indeed and moreover to know at the very outset just exactly what steps are necessary to solve the problem and solve the computer has been in a sense nothing but a very

elaborate calculating machine. But now we're making the computer be more like a almost like a human assistant and the computer will seem to have some intelligence. It doesn't really only the intelligence that we put in it but it will seem to have intelligence. In the old days to solve a problem it was necessary to be fairly expert at programming and to write out in detail on a typewriter or in punch card form all of the steps, all of the ritual that it takes to solve a problem. Because the computer is so literal minded? Because it's very literal minded. If you for example in the old days made so much as one mistake of a comma in the wrong place or a decimal point that was omitted the entire program would hang up and wouldn't run. But nowadays if you make a mistake you can correct it as you'll see immediately in the computer is much more tolerant and

much more flexible. We met next with Mr. Timothy Johnson of the Design Division of the Department of Mechanical Engineering and asked him to show us this computer and its sketchpad. We're at the TX2 console at Leckin Lab. This machine at the large computer was built by Lincoln Lab in 1956 as a research machine. How does this differ from a computer that would be used to run your bank account or something like that? Well it was designed specifically for a study of manual intervention where the man can command a computer to take different courses of action while the program is running. You can see we have several unusual pieces of input output equipment here. We have a scope, a knob, these are unusual at the time and push buttons, toggle switches. We have several other related devices. This made the TX2 a prime candidate for the sketchpad developments back in 1961 and it remained a program in this machine so it would become a coherent partner in graphics so the man can communicate with the machine.

How do you actually go about communicating with a computer in a graphical sense? Well we are using a oscilloscope here which is much like a TV set except it's being driven by the computer. In order to get the information into the computer we have to draw somehow in this display and we use the light pen. This light pen is a photo diode. It's a receiving device much like an electric eye and it will tell the computer when a single point of light has been brightened within its field of view that is seen at point. Now these letters ink here which is a brand of humor by Ivan Sullivan because what we have to do is ink up this light receiving device somehow. These letters are really a series of small points. All of them are sensitive to the light pen and if I bring the light pen such that it sees any of these points it immediately commands the computer to display a cluster of points within its field of view. So the computer knows which dot is being seen by the light pen because it knows which dot it was drawing at that moment. That's correct. Now the computer

is further know is that it must continually try to re -center this cluster of points within the field of view of the pen as we move the pen. I see a little cross there with the that's supposed to dot in the middle as the actual location we're talking about. Correct. The cross is only an expedient so we can locate the center of the pen. Now I've lost tracking there by moving the pen a little bit too fast and notice that if I point anywhere else in the screen nothing happens. I'm not squirting light out of my light pen. Right. You're picking up something. Well in order to construct a meaningful engineering drawing we have to have several graphical manipulations. Ivan's Sullivan's programs can draw straight lines and circles. Well that's about what you do in the drafting equipment anyway isn't it? It's very good to start. Alright Claire. In order to do this we can position this bright spot in the middle of the cross that you notice at a desired location and we press the button to command the computer to draw a line. It will draw a line from this position where I am now to any subsequent position of my light pen. This is much

like a rubber band stuck on two pins. One is nailed on the screen here and the other is at my light pen. So I can position this anywhere I want. Now I lost tracking there. I move the pen too fast and that told the computer to stop drawing the line. In other words any signal of any loss of tracking by a rapid flick of the pen will tell the computer I won't really mean to stop drawing the line there. I can draw other lines and what I would like to do is draw these other lines to acrylic. For instance I might want to position exactly on the end point of that line. Well if you notice that bright dot will jump onto the line as I get close to it. Well the dot in the center of the cross when you get close to the line jumps onto it. Correct. What does it do that? It's much like a gravity field at the end point. It is even a higher gravity field. It will allow us to position the point exactly on the line or in this case exactly at the end point. This allows me to move my pen quite closely. Be sloppy while I am drawing and get a precision drawing on at the same time.

So now I am going to draw a second line. There we go. And even a third one. Now in an ordinary pencil and paper drawing all we have is this particular picture. But the computer understands its geometry of the drawing here. What do I mean? I mean that if I point at this particular point and tell the computer to move that point by another push button command it will move not only that point but all three lines that are attached to it. And the delay between it's doing what you want it to is because it's computing all these changes. That's correct. Now if I made a mistake I could delete my mistake by pointing at the line in question for instance impressing the appropriate button. It's gone. Now I mentioned before we could draw circles also. In order to do this I must first indicate the center of my circle. Let's choose it to be here. And then I'll

move out to an initial radius. Let's say this point right here and I press the second button to start drawing the circle. Here's a circle. Let me reduce the drawing slice so our circle shows on the computer. You see as I move the pen it is ignoring the radial position. I've just gone off the line. It's going to be very as sloppy as you like. In other words the computer has supplied the compass here. Much like is it supplied the straight edge for the straight lines. If you go backwards you erase it. And I can wind it up the other way. Now if I tell the computer to put that point right at this circle right in that point right there the computer knows that those must be connected. It turns out that they're not really connected. It's a very small knob there. Let me move this away and show you that they're really not connected. But I have told it by terminating the point of the pen at that position.

It must be connected. Now I can tell the computer to satisfy this constraint this command by bringing in a program under command of this toggle and watch this go you see now that indeed the circle is ending at that line. We have constrained the drawing to behave this way. Now I wonder if you'd expand a little more on this idea of a constraint just what do you mean? All right let me go to a second piece of paper. What I've done here is I really say the way that drawing I just drew it. I can get it back again. I can get it back and I can select my drawing number by these toggles here. So I've selected a blank piece of paper we'll call it. We have several pieces of paper and I can let's say I'm beginning to design. I have a very nervous idea what I want to have in mind. Now as I draw my part let's say on the scope it reinforces what I have in mind. This is in general part of the design process and as I apply design criteria stresses and so on eventually I'll know what the

exact shape of this part is. I shouldn't be required to draw the exact shape to begin with at the beginning. I really don't know what it is but let's say I've decided eventually in this model that I want these to be horizontal vertical a box. I can apply a new constraint, a horizontal constraint here, a vertical constraint here and a horizontal here by pointing at the line and pressing a button. Well nothing is happening yet because remember I still must tickle that toggle over there to command the computer to satisfy these constraints. I see it won't actually let those rubber bands relax then until you say so. Right and I'll do that and watch again there we have a box. Now this idea of having constraints like this being able to make lines meet or to make them horizontal or vertical makes it quite a step ahead of something that's just a drafting machine that allows you to draw. Exactly. In other words you can get the apology down or the part and any subsequent position in your design. You can make it to behave exactly what you want. Straighten the drawing up. In other words you don't have to

draw exactly at the beginning like you have to do in drafting. This is of course just one aspect of the of the program. Now what I can do in addition to this is call up copies of master pictures. Remember that picture we drew before. I think it looks something like that. What I've done is I regard that first picture as a master. I call up copy of it and I can manipulate it locally. I can reduce it, magnify it, I can rotate it and then we place it right there and I can do this several times. This is of course very instrumental for repetitive drawings like circuit diagrams or bridge bays where we have several repetitive structures. Oh if you were drawing a circuit diagram you might have little resistors or transistors or something already drawn and stored away and you could call up as many of those as you wanted. Right. Now imagine that when I was doing my design work I made a mistake in my master. Well

in order to correct that mistake it would go back and let's say I don't really want to circular segment to be in here. I erase it. Now I have the problem of making these changes to all the currencies of this copy and my working drawing. This is very tedious nowadays with pencil paper we have to remember where all the changes are. Now you notice you remember the drawing that we now have lost our circular arcs. So I have to manufacture for instance some electronic part change the design somewhere one of these years you can just automatically change all the drawings in which it appeared. Correct. Well now I've showed you many of the basic graphical manipulations we have available. Incidentally I would like to ask you how big this piece of paper that you keep referring to is and how many pieces of paper you have available to you. Ah well this scope which measures about seven inches on the side we regard this as a window that we can move over our paper and enlarge the size of this window. We can imagine the computer as a fixed sheet of paper behind this window. It

scales approximately two miles inside. Two miles? Right and let's look at that. I can reduce this drawing slightly and let me call up a copy of that master drawing again. Put it over the center there. Oops. Now I've hit one stop already. That's as small as you can. You're looking at the whole piece of paper so. Correct. And let me magnify it now and now it's magnified so it's practically off the screen. Place another one in there. So like the picture within a picture within a picture idea. Right it's real nightmare material. It gets smaller even though the spot sort of disappears. It's really still there isn't it? Right the computer has this all memorized in its memory. It's really remarkable. Now that you've shown us some of the basic techniques for drawing and putting constraints

on can you tell us about some of the things you might be able to do with it? Well with graphics we have a very rapid way of entering information into the computer. Here I previously draw in a bridge truss with the straight lines as you saw before. Now I've done one extra thing to these lines. I want them to behave like a bridge. On a bridge we have steel members that remain of constant length really and they put loads on them. They stretch slightly. I have told the computer by supplying additional constraints which we haven't covered to make each of these lines of constant length. The numbers here indicate that if there's any change occurring in the length of these lines however my note the numbers will indicate the amount of this change. How can they change? Well if I pull this point in the middle down a slight amount, it will now stretch this bridge. It can't satisfy the constraints of the Automate of Constant length anymore so it does the best it can in a relaxation procedure and a mathematical technique this is.

To satisfy the constraint it can't quite do it and the numbers we will see will be turning over indicating the new lengths of these beams. Now with appropriate assumptions these changes and lengths can be regarded as forces. So we are now able to do a stress analysis rapidly by supplying the information graphically. Well let's do that. I'm now beginning to satisfy constraints and now I told you I wanted to move that line down slightly. Which is like pulling on the middle of the bridge oh yeah it's actually sagging. And there we are the numbers have changed and I can magnify this slightly to show you the numbers in greater detail. Here we are. Some of them seem to be negative and some of them positive. Why is that? These indicate whether the members are being stretched or compressed. The negative ones indicate the one that are being stressed. That's logical. And as I turn this now but it will stretch this point down further you can see it moving slightly and the numbers are changing indicating that more stress

has been applied to our bridge here. So really again you've gone another step further than just drafting if using the computer having drawn something to actually work out some problems to analyze this problem and actually do things with it. Correct. And we have this system has been applied to many other problems. We are drawing mechanical linkages where we can rotate the arm dynamically on the screen and watch all the associated arms which are connected to this move. So we can analyze dynamically also. You'd be able to do this without actually building a model and you'd know that if one of those numbers got too big that means that piece of steel is going to give way so let's try it either a smaller load or a different design. Well most of the things though that we live with in this world are three -dimensional rather than two -dimensional pictures like that. Is it possible to use the computer and that kind of problem? Yes we've expanded I even sell this program in the three dimensions. I have to bring that off the magnetic tape. There we have that now. Here we have a single three -dimensional object

as seen from four separate views. We have a top view is indicated by the T here a front view and a side view. Well this is the way a mechanical drawing would be laid out and they're together the other one is a perspective. Right with this addition. We can rotate this perspective separately from these three views. You get an idea of what we have here. Now begin to rotate it you see it's rotating by an axis perpendicular imaginary floor. We have a wireframe object here with no fabric covering. Hence we see the rearward lines as they come in behind this S which might be lying in the surface. But there is no fabric here so we see everything. When the letters go around behind their backwards. Right. Now we are drawing much like we are in two -dimensional except we're with that is by moving a single point around with a light pen. But we're drawing directly in 3D.

Here I have latched the pen on to the letter S and as I move this around in the letter S you see four dots moving in all four views. This is the projection of the single dot. In the side view it's actually also following the S but in the other two you're sort of looking at the S on edge so it's just moving back and forth. Right and this is because the S is indeed in the plane of that side surface. So we're seeing this single dot simultaneously in all four views as we're moving directly in three dimensions. You're just tracking in that case can you actually draw something in all three dimensions simultaneously? Let's put a rough on this object in this fashion. We'll latch on to that corner and we'll draw pyramid. You can see from the side view it's a false front house there. Right. And I can also distort things, move things like we can in two dimensions.

If I latch on to this line here or let's say this line over here I can pull that out to the side and distort the object. Now let's see what we've drawn here. We'll rotate the perspective this view again. But it's strange looking at you. And now we have a warped house in our construction perhaps. And there's the F backwards in the perspective. Tell me is it possible to do other than these simple straight line shapes. For instance curve shapes you design something like an automobile. This way. For well underway with surfaces we've begun the programming there and we understand them we think pretty well. To learn more about the handling of three -dimensional objects in the computer we met next with Dr. Lawrence Roberts of the staff of the MIT Lincoln Laboratory. Well you heard Tim Johnson explain how to construct solid objects with lines and wire figures.

However if we want to manipulate solid objects with the computer we want to be able to represent their surfaces and so on as solid. Now here we have a representation of a piece of wood perhaps which is all of the lines of the sections behind each piece are hidden. As you will see as I start to rotate this object the computer will place part of it behind the other part and you will see that indeed the computer has a representation of it which knows that it's solid. It's no longer just the wire frame it really is solid and when a line is behind the front surface you just don't see it. That's right. It somehow knows how to eliminate those lines. Well as you can see it takes the computer quite a while to figure this out. It's taking it five seconds per movement or so but this process is going on and we could indeed insert other processes like gravity and solidarity of objects in other ways if we wish to. Now how do you draw something like this and which view do you start with?

Well we don't draw it with any view. We start by using projections of particular objects we have in mind. If I start over by clearing out this picture and now I will insert just a single and instance of a cube. This is just one picture. Now this doesn't look like much yet but you will see that the computer knows that it is solid. It's an object and I can take and compress this in various dimensions and make it into any transformation of a cube that I wish and I can move it around until it's in the position that I wish to have it in the view. And so this is just one of my models that I can work with. I have a few others that I can call up for instance a wedge. I'll make the wedge slightly smaller and

rotate it a little bit so you can see the properties of it. Now these will be fairly basic shapes. Oh yes you can construct quite a bit out of these basic shapes. As I move the wedge around in space you will see that it goes behind the solid and through it. And the computer figures out where the line should appear and where they should be. Where the intersection is. You get the feeling of three -dimensional space here very dramatically as though this were a window and it was a four area and a behind area. Right now this comes out through the other object you see that it indeed intersects it and can move right through it. Now aside from these problems in mechanical design and the analysis of things like the bridge structure what other problems are suitable for using this technique on? Well we have a lot of areas in which we use we try to use graphical communication with the computer because it is the natural language to talk about your problem. For instance in electronic

circuit design the engineer would like to draw his circuit and then simulate it to see how the circuit works under some set of conditions. And so we set up a program with SketchBet so he can draw it and simulate the action of the circuit and watch the waveforms on the computer. Now we also have been working with the flow charting of programs. To instruct a computer what to do you need to write a program. A flow chart then would be sort of a diagram of the steps that you'd want the computer to take in solving some particular problem. Yes in fact I have a flow chart on here. This is with SketchBet again and we have a demonstration of boxes representing statements to the computer to do some operation and compare some numbers and make a test and transfer one way or the other. This is the way the human being would like to set it up by drawing boxes like this would represent different computations. This is the way that a programmer normally operates and then he has to transcribe this to some form like cards or something as an input. But here we go directly from drawing the flow chart and

stating what each piece is putting the statements inside the boxes to a compiled program which he may execute on the computer. Well you've shown us how you can enter numbers say by a typewriter and now we can draw things for the computer directly. Are there other ways that we'd like to be able to communicate with the computer? Well we've been investigating here at the lab speech recognition and handwriting recognition for enabling the person to communicate better in those ways with the computer. There are a lot of techniques which would be very useful when combined into a whole and all of these techniques including the graphical manipulation will make it much easier in the future for the man to dynamically converse with a computer. Well thank you very much Dr. Roberts. Today we've been visiting the MIT Lincoln Laboratory to learn about the computer sketchpad. Our guests have been Professor Stephen Coons, Mr. Timothy Johnson and Dr. Lawrence Roberts. I'm John Fitch, MIT Science Reporter. This is NET,

National Educational Television.