How are the computer and the robot affecting the way we work? NOVA chronicles the new industrial revolution reshaping the American workplace.
Category
📺
TVTranscript
00:00Since the Industrial Revolution, new technology has helped American manufacturing grow into
00:17world leadership.
00:21But today, many industries have lost their competitive edge.
00:29Can the new technology of advanced automation turn things around?
00:34What is this new technology?
00:36How does it work?
00:39NOVA examines the robot revolution and the future of the American workplace.
00:51Major funding for NOVA is provided by this station and other public television stations
00:55nationwide.
00:58Additional funding was provided by the Johnson & Johnson family of companies, supplying health
01:02care products worldwide.
01:08And Allied Corporation, a world leader in advanced technology products for the aerospace,
01:12and automotive, chemicals, and electronics industries.
01:32Through the 19th century, America industrialized.
01:52New machines expanded the productivity of workers.
01:56America developed into one of the world's leading manufacturing countries.
02:02With the 20th century came new technology, the machines of mass production and the assembly
02:07line.
02:11This country grew to lead all others, both in its productivity and prosperity.
02:23Yet today, certain sectors of American industry are bankrupt.
02:30In the last decade, two million manufacturing jobs have been lost.
02:35More than 1,000 companies have closed.
02:38America, once maker of half of the world's steel, today imports a third of her needs.
02:46Some blame management or the unions.
02:49Others blame obsolete manufacturing methods and equipment.
02:54The basic moral of the story is it's the fault of everybody.
02:57You can't build a world-class economy without world-class inputs, and if you look at any
03:02of the inputs going to American industry, the skills of the workforce, the qualities
03:07of the managers, the amount of capital equipment, the research and development, American industry
03:11doesn't shape up on any of those dimensions.
03:15And then there's foreign competition.
03:17Today, the United States imports 25% of her cars, 40% of her vacuum cleaners, 100% of
03:25her videocassette recorders.
03:30Foreign productivity and ingenuity are challenging American preeminence in mass production.
03:35I think that the real problem in the competitive area at the moment is the United States is
03:39in that denial phase.
03:41We see ourselves being beaten up in international competition, and we jump to the conclusion
03:45that somehow the rest of the world is doing it by cheating.
03:47The Japanese are being unfair.
03:49There's nothing unfair about a cheap, high-quality product.
03:52That's what the game is all about.
03:56The American economy is in transition.
03:58Old industries are dying.
04:01Foreign products, cheaper and of better quality, are taking over American markets.
04:08Yet revolutionary forces are on the move, heralding a promise of rejuvenation, the forces
04:14of computer-driven automation, ranging from computer-aided design to computer-programmable
04:26robots, computers manipulating information, computers controlling machines.
04:40Is this new computer technology the means for America to revitalize, to produce better
04:48quality goods at lower costs than countries abroad?
04:53We have over 500 computers supporting this manufacturing process in this plant.
04:59Four years ago in our company, we had less than a dozen computers supporting manufacturing
05:05in its daily production of its goods.
05:07That is the explosion of science.
05:13How far can computer automation deliver the benefits to America of greater productivity,
05:20improved quality, reduced labor costs, and more manufacturing flexibility?
05:29But yet, how far could computer automation change and disrupt our society?
05:36Obviously change is inevitable.
05:37I think obviously we'll see a lot more computer technology.
05:40The real question is, what is the character of that technology going to be?
05:44How are these machines and systems going to be designed, and what will the role of people
05:49be in the workplace of the future?
05:53This program will examine the technology of computer automation, how far it's being introduced,
05:58and the choices it presents both to industry and society.
06:06The choices between different systems and how they might be developed to reap the greatest
06:11economic benefits with the least social cost.
06:15We have an unprecedented array of choices before us.
06:19We have an unprecedented potential to use computers and computer-based automation in
06:25a way that benefits the society as a whole.
06:27I think it would be tragic not to seize that potential and use these technologies in a
06:32way that in fact do bring real benefits.
06:38Already computer automation is appearing in offices and factories, and the key element
06:43in this new automation is the computer.
06:48A computer, simply, is a device which electronically stores information or data.
06:54The data might be numbers, words, or images.
07:00This data can be manipulated according to a set of mathematical rules called a program
07:05or software.
07:08A program can multiply numbers, generate text, alter images, or send electronic commands
07:20to machines, all processes previously done by human hands and minds.
07:29Computers first appeared more than 40 years ago.
07:32Since then, they've been getting cheaper, more efficient, and much more powerful.
07:39By the early 70s, many organizations had installed a computer, probably in the basement, to help
07:45with financial or personnel records.
07:49These computers were cost-effective for large bureaucracies.
07:52Still, they had little impact on the way most Americans worked.
07:59Then in the mid-1970s came the microprocessor, a whole computer packed into a tiny chip,
08:06and computers passed a watershed.
08:11Now they were small enough, cheap enough, and powerful enough for a host of new applications,
08:17which might help revitalize American manufacturing in the nick of time.
08:23In the second part of the 1970s, we began to hear about the fact that other countries
08:28were beginning to be as productive as we were.
08:30We began to see many foreign goods coming into the country, especially electronics,
08:34automobiles, so on.
08:36This was perceived as a threat, and people looked around for some response to this threat.
08:40Well, the microprocessor had just become fairly readily available, and so people reasoned
08:46that we could possibly bring this computer power to bear in the manufacturing environment
08:51by having these microprocessors direct machines to do things that they could never do before.
08:56We began to look into having microprocessors direct robots, which could conceivably work
09:01day and night producing products, and this was going to help us with our productivity
09:04problems.
09:07Robots are computer-programmable mechanical devices.
09:13These fluid-moving robots are symbols of the extent of computer power.
09:17They mimic human actions.
09:23By the mid-1980s, computer-controlled robots needing no coffee breaks, safety regulations,
09:29or unions were replacing some of America's 20 million workers.
09:35This type of robot was hyped by industry and the media as the great savior of manufacturing.
09:43But there are only 15,000 robots in American factories today, directly displacing perhaps
09:5020,000 workers.
09:56Most robots are in showcase factories of large companies, like General Motors, John Deere,
10:01IBM, Chrysler, or General Electric, whose managements are willing and able to make investments
10:07of many hundreds of millions of dollars.
10:14However, such robots have shown limited cost-effectiveness.
10:21They're used primarily in spot welding and paint spraying, where protection against hazards
10:26make it particularly expensive to use human workers.
10:34But robots are not, of themselves, the key to computer automation.
10:39They're only a peripheral factor, because the great advantage of computer automation
10:44centers on the computer.
10:50The computer can not only control machines, but it could also control the organization
10:55of a factory.
10:57Organization control is where the greatest benefits of computer automation could be achieved.
11:09Already, the computer has started to revolutionize the control of office and service organizations.
11:15Here at the New York Stock Exchange, productivity is up 400 percent, because the computer network
11:22electronically stores, tracks, and records data.
11:28Before computer automation, the paper-bound records could lag far behind on even a quarter
11:33of present-day business.
11:38Today, more than 10 million computer terminals are used in offices across America, from banks
11:46to insurance companies. They track, store, and manipulate data electronically.
11:54Two out of every three office workers use some form of computer-based machine at some
11:59time during the day to help them streamline their work.
12:06In Boston, for example, they've installed an $18 million computer system to help run
12:12the city electronically.
12:14It's a very old city. We have to maintain birth records, assessing records, deeds records
12:19since 1629. For that public information to be available to either historians or people
12:24who need to utilize that in their daily work, it has to be readily available. And right
12:30now, there isn't enough physical space in the building to actually support that kind
12:36of public information.
12:39The weights and measures department at City Hall is a holdout, not yet computerized. It's
12:44paper-bound. Here, clerks must still dig out information buried in banks of files and hand
12:51process it.
12:59This is the city mainframe computer. It electronically stores almost all city records and data.
13:07The mainframe is the center of the Boston Automation Network. It links every city department
13:12and the school, hospital, and police offices. Data can be channeled electronically between
13:19a dozen different departments and can be punched up in seconds.
13:26For example, before computer automation, Boston had had to declare a moratorium on unpaid
13:31parking tickets. They couldn't deal with the paperwork.
13:36Today, at any city terminal, they can compare the tax records with public works needs, or
13:43birth records with school enrollments, and much more.
13:49The Boston City computer system replaces paper handling with electronic connections. Each
13:56department can have immediate access to all others, and to the mainframe data store. It
14:03allows this large organization to be controlled more efficiently.
14:09In the same way, computers can boost control efficiency outside the office, in the much
14:15more complex field of factory manufacturing. Factory manufacturing is a complex of separated
14:22sectors of work, divided by barriers of understanding and working methods. There are the sectors
14:30of design, testing, fabrication, assembly, inspection, and materials handling. People
14:38from clerks to managers, doing reams of paperwork, bridge the sectors and tie the operation together.
14:57Computer automation is appearing independently within each sector, varying from one sector
15:07from rudimentary to advanced. For example, this is a computer-aided design terminal.
15:15The computer stores and manipulates design data, eliminating the hand-drawn design.
15:24This is an example of computer-controlled assembly. The computer sends command data
15:29to robots spot welding cars on the assembly line.
15:35This is a computer-automated materials handling system. The computer processes data to direct
15:43automated carts moving materials on the factory floor.
15:49Manufacturing is a very complex activity, and we've tended to deal with that complexity
15:54by breaking it down into smaller, more manageable parts. We develop these islands within the
16:02factory environment, and we automate those islands, but the integration between those
16:07islands has yet to happen.
16:11Despite computer automation within sectors, white-collar workers must still do the paperwork,
16:18passing information along to keep the factory running and deal with breakdowns and disruptions.
16:25That's pretty much where many companies are today, with islands of automation, and it's
16:30the integration of those islands that really will bring the productivity gains in the future.
16:38Automation experts foresee the greatest productivity gains coming from replacing people and paper
16:44with computer control. Computer-integrated manufacturing. SIM. Computers linking every
16:52factory sector together. Computers monitoring for breakdowns and disruptions to create a
16:58workerless, paperless factory.
17:02We will now investigate the extent of computer automation within each factory sector and
17:08the possibilities of computer links between them. We start with design.
17:19Design drawing has long been the basis for manufacturing, but like the old-fashioned
17:25office, it's paper-bound. Unlike electronic computer data, it can't be manipulated, processed
17:32and analyzed in fractions of seconds. Drawing is labor-intensive, and it's beginning to
17:39become a thing of the past.
17:44This is a computer-aided design terminal, a CAD terminal, a tool to automate design
17:50work. Instead of a pencil, the designer uses a light pen to pull lines and curves out of
17:57the computer. He's constructing a spindle for a car disc brake.
18:06He can also pull previously designed parts out of the computer memory. Using this tool,
18:12there's no need to search through blueprint libraries or transcribe what has been done
18:17before. No need to reinvent the wheel.
18:22At a touch, a wire mesh outline becomes solid. It becomes a cutaway, a blow-up. An assembly
18:36explodes into its parts. Within a solid image lies its wire mesh foundation. In reality,
18:44computer design isn't as simple as in this idealized demonstration, which took many hours
18:52of setting up. Nevertheless, it's used extensively.
18:57Computer-aided design systems are used extensively today, from the very largest of companies
19:01even to small companies. One reason for that is that computers themselves started in an
19:08office environment many, many years ago. It started with things like payroll and accounting,
19:13and then gradually got into the technical side, got into the engineering and design,
19:18and only now are we starting to go into the manufacturing side of the operations.
19:23At General Motors, a leader in automation technology, they're using more than $100 million
19:29of CAD systems for original car design work. A typical car body involves over 5,000 elements.
19:39Using CAD, these elements can be interrelated, adding to the exterior framework interior
19:47items from the firewall to the steering column. Design items can be tried out in seconds.
19:57CAD systems generally bring to companies productivities typically three, four, or five to one, and
20:04these represent real savings in terms of the time it takes to generate a design, the amount
20:12of work, the amount of designs that can be done in a given period of time by designers.
20:17So these are the kinds of gains that we see. Three, four, five to one.
20:23Whatever the benefits of computer design, it would seem that manual drafting will become
20:28as anachronistic as the slide rule in this phase of manufacturing.
20:37The questions of productivity and jobs are also reflected in the next stage of manufacturing
20:42automation, testing. Road testing a car, for example, might become a thing of the past.
20:49The data from these tests might, in the future, be derived from a computer engineering simulation.
21:05This engineer is simulating the mechanical testing of designs and analyzing structural
21:10stresses. She's examining a version of the real-world behavior of a car.
21:19She's analyzing the behavior of a product before it's built.
21:25This is a stress analysis of a wheel. Through an overlay mesh, the stresses are analyzed
21:31in each tiny segment to reflect the distortions in a wheel running over rough ground.
21:40The engineer at the computer looks for the effects of vibrations, friction, and fundamental
21:46frequencies. The engineer at the computer could replace the test driver.
21:55But many aspects of computer testing are experimental. This sector of automation is far from developed.
22:05The computer would have to know all about every possible stress a car might suffer on
22:09the road to fully automate this stage of manufacturing.
22:20After design, a product is fabricated. Fabrication is the next step in the industrial cycle.
22:26It takes us to the factory floor, where machines cut and shape parts to make, for example,
22:32some of the 5,000 components of an automobile. Programmable automation to control these processes
22:39started decades ago.
22:43It's interesting that about 30 years ago, when this kind of automation began, the machine
22:48instructions were stored on this kind of paper tape. So it would be the machinist that was
22:54essentially responsible for taking the design data and making this piano roll kind of tape,
23:01which gives both the geometrical positions of the component, and it also governs the
23:07machine speeds and the table speeds, the speed of the machine.
23:12Nowadays, we don't need to use this older style. We can program the information directly
23:18on the console here on the computer.
23:24But the programming still depends on the machinist, who might have to be retrained for this job.
23:31He uses his knowledge and experience of metalworking and tools to translate design data into machine
23:38commands.
23:47The John Deere tractor plant in Iowa is a showcase for such computer-automated fabrication.
23:55It represents what can be achieved with today's technology, $500 million, and a management
24:03which is willing to take the risk.
24:07But this automated machining floor is one of only 30 such installations in America.
24:15Here six computer-programmable machining cells shape engine blocks. They change their parts
24:22to do a multitude of different jobs, one after the other, untouched by human hands, controlled
24:32by computers.
24:35They save in setting up time. There is less scrap. The quality is more consistent. Inventory
24:44can be cut by two-thirds, lead time by half.
24:50The overall gain in productivity, Deere says, has been hard to quantify, but systems like
24:56these generally improve productivity 20 to 30 percent.
25:02And most importantly, they can quickly change from making one type of part to making a different
25:08type of part. This flexibility is the big bonus of computer automation. A whole variety
25:15of custom products can be made at mass production prices.
25:20Previously, before the advent of the computer, that manufacturer, if he wanted to change
25:28products or change his machine, had to do it mechanically. He had to change fixtures
25:32and tools and make these mechanical changes. What we see now, and certainly even to a greater
25:38degree in the future, are these changes coming about by changes in the software, by reprogramming
25:44rather than refixturing and retooling.
25:48Yet reprogramming still calls for the machinist craftsman to use his knowledge and expertise
25:54to translate design blueprints into computer commands. One step toward the factory of the
26:04future would be to incorporate the craftsman's expertise into a computer.
26:11The craftsman understands about the sloppiness of the machine tool, or how the tools will
26:17deflect, or how the fixtures might move whilst the component's being machined. And he does
26:23things like look at the chip color, and he looks at the way in which the surface finish
26:28on the component is being formed. Now, once we begin to get some of that data into the
26:34machine, then we'll be able to do this hands-off thing much easier, and we'll begin to move
26:40towards what you might call an expert system for machining. Now, an expert system is a
26:45computer program which contains all of the geometrical knowledge and all of the material
26:49knowledge, plus all of this craftsman knowledge, and it can begin to do some thinking or some
26:55inferences about the machining environment and the manufacturing environment. But once
27:00we have all of those rules of thumb that these craftsmen have built up for many years, also
27:05embodied in the electronic form, then we'll be able to do this unmanned machining much
27:10more easily.
27:13While such expert systems are not generally available as yet, on this factory floor the
27:19traditional machinist no longer has any place. Only about 10% of machining is computer controlled
27:28in the United States, but it might already account for as much as 50% of the value of
27:33all machining work. It means higher productivity and fewer employees. And though numbers are
27:48difficult to assess, it's easy to see why workers are worried. Union leaders consider
27:54this a major issue.
27:55We're going to see a drastic change. The jobs that we have now are going to become as obsolete,
28:02as a matter of fact, 15, 20 years from now, somebody's going to look and see a picture
28:06of how we work, and it's going to look as obsolete to them. It could be your children
28:11or grandchildren. It's going to look as obsolete to them as a horse and wagon looks to us.
28:19But how many people out here actually feel that this is a problem of the future and that
28:23our jobs are being threatened in the very near future? Raise your hand.
28:30Among many conflicting reports, the Bureau of Labor Statistics projects a shift from
28:35high-paid manufacturing to low-paid service work over the next decade. With 4 million
28:42manufacturing jobs lost due to automation and other factors, while 10 million service
28:48jobs are created.
28:49We could, in fact, see a very likely scenario where the society becomes increasingly polarized,
28:56where many of the best industrial jobs, for example, are eliminated, and the only jobs
29:00that remain are those jobs that it's not worth automating, low-paid, dead-end, short-work
29:06time jobs. If the technology is going to victimize people, they're going to fight against it.
29:11They're not fighting against technology. They're fighting against something that victimizes
29:15them.
29:16I'm not discouraged by computers. I'm threatened by them.
29:20But like Angie said a little while ago, hey, you look at that computer and you say, you're
29:25my mortgage payment. You're my car payment, right? I'm going to beat you and I'm going
29:30to learn how to run you. If it means my job, I'm going to do it. But for God's sakes, give
29:34us the opportunity to learn anyway.
29:36People simply have to be retrained and reassigned. It is what I would call, sir, a relocation
29:42of people and a re-understanding of what their function is.
29:47As long as they could continue to give me job security, then I would be happy. I don't
29:53want anybody to tell me, Bob, this computer is going to do your job. It's been nice. Good
29:58luck.
30:01Some large companies, like General Electric, have retraining programs. But like automation
30:06itself, this is the exception rather than the rule.
30:09It went from tubes into transistors and diodes. At that time, it was a little more difficult.
30:16Retraining concerns tend to come after the fact rather than before. In America, unlike
30:22other countries, there's no national strategy and little government help for these efforts.
30:29But even the strongest advocates of automation believe that at the company level, consultation
30:35between management and labor could ease the introduction of computer automation.
30:40I think that the planning process that involves all of the parties concerned, senior management,
30:45middle management, the working level, should take place as early as possible. Particularly
30:52the working level people should be brought in early so that they can see the computer
30:57automation not as a threat to their jobs, but indeed as a way for that company to more
31:03effectively compete in the marketplace and, in fact, to create more jobs by growth of
31:08the company and by producing better quality products at lower cost.
31:12However, computer automation has barely touched American industry. Many companies have balked
31:21at the capital expense. There are still 20 million factory workers and only 15,000 robots.
31:30Even the leading companies, the Fortune 500 manufacturing companies, have applied computer
31:37automation no more than 5 or 10 percent of the potential that they can apply it.
31:47And in assembly work, the next sector of the manufacturing cycle, at this John Deere showcase
31:53automation plant, they haven't brought in robots to replace assembly line workers. Why?
32:00To find out, we must look at computer-controlled robots and understand their abilities and
32:06their limits in replacing human beings.
32:11The simplest kind of programmable robot is a point-to-point machine. It'll be programmed
32:17to go from one fixed X, Y position, say, along a track to another fixed X, Y position,
32:23and it will be doing a pick-and-place operation during that motion.
32:30This is a computer-controlled pick-and-place assembly robot with three degrees of freedom.
32:37Its movements are simple and it can be reprogrammed to work in different configurations.
32:45The Japanese have made these simple robots their main investment. Seventy percent of
32:50their 50,000 robots are of this type. Only 40 percent of America's 15,000 robots are
32:57as simple as these.
33:01At the next level up, we have a machine which has a greater number of degrees of freedom
33:06in the overall working system. For example, the base, which is rather like the human shoulder,
33:15the elbow, rather like the human elbow, and these wrist joints here enable this machine
33:22to obtain six degrees of freedom. And so this can go over a more complex path and do things
33:28like spray painting and spot welding.
33:33These are computer-controlled robots with six complex joints designed to imitate human
33:38arms. They inherit the same limitations. They move no faster than human arms. They
33:48are not precisely accurate, and they're limited in the loads they can carry. And if an error
33:55puts the work in the wrong place, they never know it. They continue on mindlessly, but
34:02they don't do their job.
34:07Another important factor is that robots are much more expensive than people in general.
34:11A robot will cost about $50,000, maybe more. It'll take another $100,000 in order to install
34:18it. And if you're lucky, it will do the job that one person will do. Now, if you're talking
34:23about competing against someone in a foreign country who's earning maybe $4,000 or $5,000
34:29a year at the most, you can see that the robot will be worn out long before it pays for itself.
34:35But regardless of cost, are human-like robots the best way to automate assembly work? What
34:41tasks would they have to be able to do?
34:43For our tasks, it's very easy for me to do something like putting a nut on a bolt. It
34:49would be very difficult for a robot to do. In order for me to do it, I need to be able
34:53to sense the contact between the nut and the bolt. I can tell from the feedback I get through
35:00my fingers whether the nut is the right size for the bolt, whether the threads are crossed.
35:04For a robot to do a job like this, it would have to process a very large amount of sensed
35:08information. It would have to sense this contact. It would have to be able to sense the torque
35:12so that it knew the threads weren't crossed. It would have to allow the nut to move down
35:16along the bolt while it was turning it. One of the reasons that people study this is because
35:20they're interested in how they do these tasks. And by studying how a robot might do them,
35:24it gives them some insights into how people work.
35:26There's an important distinction, though. We have to realize that this isn't necessarily
35:30going to be important for manufacturing. Duplicating what humans can do isn't going
35:35to be terribly productive in the manufacturing environment because we already have humans
35:38that can do it.
35:41Indeed, in Detroit, human welders still work in parallel with robots when awkwardly placed
35:47or specialized welding is needed. It's difficult to conceive of a cost-effective robot that
35:53could connect cables on this assembly line.
35:58The idea that robots should be complex has grown directly out of the fact that most of
36:04the American robots developed through artificial intelligence laboratories where the goal was
36:08to mimic human behavior. In fact, human behavior is not particularly what you want to mimic
36:14in a factory.
36:17Nevertheless, certain factory applications will come out of university laboratories.
36:23At the University of Utah, scientists are building the kind of humanoid robot which
36:28Professor Searing wouldn't advocate for today's factory. They work in artificial intelligence,
36:35trying to understand how humans function by building a hand which mimics the dexterity
36:40of a human hand. How is it constructed?
36:43Well, it essentially has to have the four subsystems that we talk about within this
36:48dexterous hand. It has to have structures, which are bones, tendons and muscles, which
36:52are the actuators, sensors to detect angles of joints and the contact with objects, and
36:58a central control system to manage the whole system.
37:03In short, to make a useful robot hand, you have to make most parts of a person, including
37:09the brain. But concentrating on the hand, these are the muscle tendons. They run over
37:17a complex of pulleys, pulled by tiny hydraulic motors, to move each finger.
37:24Fiber optics measure the joint positions for feedback to the control center, which governs
37:30the speed and the stress on each tendon, as well as coordinating all the fingers so that
37:36the hand might do a useful task rather than move spastically.
37:42But what objectives can such pure research have?
37:46It has two objectives only. It shouldn't be judged as a manufacturing device because
37:51its first goal is, as a research tool for people in artificial intelligence and other
37:56areas of robotics, to begin exploring dexterity while supervising a device that's quite rich
38:03in its degrees of freedom and sensory capabilities.
38:07I think it's important to recognize that the universities have an important role to
38:12play in the area of robotics and computer automation, and that role is properly at the
38:19leading edge of the technology.
38:24So while research looks to the future, today on these assembly lines, human hands and brains
38:31are likely to remain more cost-effective than any complex, expensive robot.
38:42A more effective approach to assembly automation is, where possible, to change the product.
38:48Redesigning the products which we now assemble manually so that they're easier to manufacture
38:53is much more important than building robots that have all the capabilities of the human being.
38:59In Lexington, Kentucky, IBM has set up a $350 million showcase automation plant.
39:06They've redesigned their new electronic typewriters from scratch so that they can be
39:11assembled using simple robots.
39:15The parts to be assembled are pre-sorted. The robots need simply pick and place.
39:22We're seeing a lot of companies put a great deal of effort into both designing new parts
39:29as well as redesigning old parts so that simple, pick and place robots can be used in the
39:36manufacturing of those parts, rather than the more sophisticated and complex machines
39:41that are just expensive to bring in and difficult to support.
39:47Traditionally, products have been designed for efficient function and human assembly.
39:53It's a radical departure to design both for efficient function and for simple robot assembly.
40:01These robots are computer controlled. They can be easily reprogrammed to assemble
40:06updated versions of this typewriter or totally new products.
40:11They bring automated flexibility into assembly.
40:17Certain jobs, like assembly of the typewriter circuit boards, can be done with the speed
40:22and accuracy which no person could match.
40:25This brings to the manufacturer advantages of better quality, reliability, productivity, and less labor.
40:35This factory is at the forefront of assembly automation in America.
40:41At present, it's unique. It's an indication that in industrial automation,
40:46redesigning the product for simple, cost-effective robot assembly is a productive way to go,
40:52that simple can be beautiful.
40:56That, though, can't be the total answer because many products are difficult to assemble.
41:01That, though, can't be the total answer because many products are difficult to redesign.
41:06For example, some experts say that car assembly work of this type
41:10might best be left unautomated for humans to do.
41:17The big advantage that humans have in a manufacturing environment
41:20is that they can alter their performance to take into account changes in the environment.
41:26For example, if parts don't fit quite right, they can change the way they try to put them together.
41:31The state of the art in robotics now is that robots can't come close to that sort of adaptive performance.
41:38But advocates of all-out automation disagree,
41:42arguing a need for complex robots to replace people, even on tricky assembly jobs like these.
41:50I think we'll see a wide spectrum of sophistication in the assembly operations.
41:55Some jobs will be able to be done by relatively unsophisticated, what we call pick-and-place robots,
42:02while other assembly tasks are going to be more difficult and more complex,
42:07and they will require the more sophisticated robot technology.
42:13For robots to do assembly work like these people,
42:16they would first need eyes to pick out the right pieces for the job.
42:21Vision is part of the more sophisticated robot technology.
42:25How would it work, and is it worth it?
42:31A human eye can easily recognize and pick out the right assembly part from this tray.
42:38But to the computer eye, the video screen,
42:41these are a jumble of overlapping shapes, difficult to analyze and recognize.
42:46Much more research into computer vision is needed to bridge this recognition gap.
42:53To analyze a scene consisting of one part is relatively simple.
42:57However, if you have a scene consisting of several parts,
43:00especially if they touch or overlap, such as this one here, where you have parts on a tray,
43:06that scene is more complex, and it's much more difficult for the computer to discern that scene.
43:13When we see a circle, or a circular object, or a spherical object, or a planar object,
43:19somehow our mind processes it on the basis of what we have grown up with,
43:27some kind of models that are built into our brain.
43:30We don't quite understand how we do it.
43:34That's the mystery of nature.
43:37That's the mystery of nature.
43:40Now, we are trying to emulate it, thinking because it's easy for us to do it,
43:45it should be easy for a computer. Ain't so.
43:50Successful computer recognition of jumbled assembly parts still has a long way to go.
43:57But computer recognition of simple, distinct shapes
44:00is already being used in the next sector of manufacturing, inspection.
44:07Back here at IBM, they're inspecting their new typewriters with a simple computer vision system.
44:13The required recognition is limited to the letters of the alphabet and type symbols.
44:18It's designed to work simply,
44:20part of the whole approach of redesigning the product for simple, programmable automation.
44:26However, in university research, scientists are working to duplicate the stereo depth vision,
44:40which humans use in inspection work.
44:43If you take a scene in which two objects are separated in depth,
44:48and you look at them first with one eye, and then with the other eye,
44:52you detect differences in those images.
44:54We're capable of matching up those differences,
44:57and on the basis of that, computing depth in the scene.
45:00When we've got the depth data from stereo or whatever,
45:03our problems are only just beginning,
45:06because what we really want to do is to compute symbolic descriptions of objects inside the computer.
45:12For example, we'd like to be able to compute a mathematical model of an object such as this coffee cup.
45:17We'd like to say that it has a body which is cylindrical, from which we drink,
45:22and that it has a handle which is attached in two places to the body,
45:25and that the handle is curved and essentially cylindrical.
45:29We'd like to have those kinds of models and store them directly in a computer,
45:34and have subsequent instances of a cup matched up against those stored models.
45:40In this way, a computer inspection system might, at very great expense, detect a chipped cup and reject it.
45:49Unfortunately, systems like I've described with a cup are only available right now in a laboratory.
45:54To compute descriptions right now, even on simple objects,
45:58requires 45 minutes to an hour of processing, even on quite powerful computers.
46:04We in this laboratory are building a computer that consists of a quarter of a million processors
46:08that are all interlinked,
46:10and which we hope that we'll be able to compute stereo in around a hundredth of a second.
46:16This would need the development of the next generation of multi-million dollar supercomputers,
46:22which are hardly likely to be cost-effective for inspecting cups.
46:28How far, then, is sophisticated technology needed to attain the greatest benefits from automation today?
46:36Technology, supercomputer or otherwise,
46:39is not the primary impediment to the introduction of computer automation to manufacturing plant today.
46:46The main impediment is management's unwillingness to take risks
46:52and to chance doing things significantly different than they've been doing them in the past.
46:59At John Deere, they have taken such a risk,
47:02with a $30 million computer-automated materials handling system.
47:09Materials handling is the next sector of manufacturing.
47:13It deals with delivering all the parts and materials to the right place in the factory at the right time.
47:21In this half-mile by half-mile factory, the system is computerized.
47:29The parts start in one of three high-rise storage areas.
47:33There are 3,000 different parts in this one.
47:37The computer knows where each part is. It directed it there in the first place.
47:44When the production schedule in the computer calls for particular parts for a particular tractor model,
47:50these robot carts go into the store, pull the parts out,
47:55and route themselves through the plant conveyors to arrive at the right place on the assembly line just in time.
48:04This flexible system can reprogram the materials flow for changing production demands.
48:11It's the showcase of the just 90 such systems working in this country.
48:16It cuts inventory by two-thirds, saving 10% of the production costs.
48:23Productivity is up because storing, moving, and tracking is controlled by computers and machines,
48:30not by manual or white-collar clerical and scheduling workers.
48:35In this sector of automation, traditional white-collar jobs disappear.
48:42Many of the people whose job it is now to keep track of the factory will no longer have anything to do.
48:48If you can cut out a white-collar job, you save substantially more money than if you cut out a blue-collar job in general.
48:54Anyone in the white-collar labor force who handles paper for a living has a job that's very vulnerable
48:59because handling paper is precisely what the computer can do away with.
49:04Computer automation, we have seen, is being brought into each sector of manufacturing.
49:10But the paper shuffling needed to run the whole system is still a barrier to a revolutionary increase in productivity.
49:18This barrier could, in theory, be overcome with the advent of computer-integrated manufacturing,
49:25a central computer replacing people and paper.
49:30However, such a central computer would need to embody the human understanding of how a factory system actually worked
49:37and be programmed to make decisions about what to do in any situation.
49:44This is exactly what Professor Mark Fox of Carnegie Mellon University is trying to set up at his computer,
49:51an artificial intelligence expert system to run a factory.
49:57But first, at his fingertips, he must have the understanding of how a factory works.
50:02Such human expertise and the responsibility for doing the job belongs to the factory production manager.
50:11In the factory, Professor Fox taps the production manager's expertise.
50:16Because running a factory is a shoot-from-the-hip affair,
50:19nobody really knows logically how to coordinate all the machines, materials, and people,
50:25and their disruptions, breakdowns, and bottlenecks.
50:30We've had a lot of problems this year getting the output to match our demand.
50:34It creates a couple problems for us and a couple concerns.
50:36One concern is feeding the bottleneck.
50:38Since it is a bottleneck and it's constraining the total output of the system,
50:41we want to maximize that output as much as possible.
50:44The second thing we have to consider...
50:46If an understanding of how a factory works could be put into a central computer,
50:51this computer could respond to and deal with breakdowns and disruptions far more quickly than any person.
51:01Manufacturing efficiency could then, perhaps,
51:04take its great leap forward to computer-integrated manufacturing.
51:08SIM.
51:14SIM represents to many, many companies a brave new world of manufacturing.
51:20It's a new technology, a new way of doing things.
51:24It involves changes in an organization's culture and in the organization's structure.
51:29I think one has to be very guarded in their expectations for the application of techniques of this nature,
51:35expert systems techniques, artificial intelligence techniques,
51:39guarded in the sense that they're just making it out of the laboratory.
51:42So if one expects that these techniques can be just picked up and used directly in the factory,
51:47then you're mistaken.
51:48There still is a great deal of development work that has to be put into these systems.
51:53Developing computer-integrated manufacturing depends on getting a better understanding
51:59of how a factory could optimally work,
52:02and how that ideal could be translated through software into a central computer control system.
52:09Computer-integrated manufacturing, if it worked,
52:13might be the technological edge to boost productivity
52:17and make this nation's industry competitive with any other country.
52:23Yet today, computer-integrated manufacturing is only a theory.
52:28Of available computer automation, there are only a few showcase examples of excellence,
52:34representing little more than 5 or 10 percent of what could be possible.
52:39While there appears to be a trend toward computer automation,
52:43American industry has hardly started to invest in the basic technology.
52:48Well-publicized showcase factories in the companies like John Deere and General Motors and IBM
52:54that are recognized as the leaders in bringing these showcase factories up to speed and using them,
52:59these showcase factories are not indicative of the rest of the country.
53:05Other companies in the United States can look to these factories as goals toward which they can strive.
53:16According to this view, America faces an urgent and major program of re-industrialization
53:22based on computer technology and the redesign of products, equipment, and attitudes.
53:30But others welcome delay, arguing that advanced automation should proceed slowly
53:36to allow time for planning the transition.
53:39What we have done here is secure people's jobs.
53:42Prior to going into heavy sophistication of automation in Chrysler,
53:46we were laying people off by the thousands.
53:49In the last 36 months where our heavy automation program has bit,
53:53we have now been bringing back well over a thousand people a month for over 24 straight months.
53:58Computers can play a vital role.
54:00Automation can be central to this economy and used in a very beneficial way.
54:05But none of that's going to happen if we focus on all technological change as inevitably meaning progress.
54:12The future is not something that we can worry about in a few years.
54:15The future is upon us.
54:17The question is what that future is going to look like and what the human cost is going to be.
54:26The availability of computer automation raises vital questions for American industry and society.
54:33Is computer automation really the key to revitalizing manufacturing?
54:38And if so, is it being adequately pursued?
54:43Certainly leading companies plan to install further automation systems.
54:48Automation seems to be working for them and others are expected to follow.
54:54Yet are traditional market forces sufficient to drive enough companies toward this next industrial revolution?
55:01Or are greater government incentives called for?
55:05While studies disagree on the actual figures, substantial numbers of workers are likely to be dislocated.
55:12Will laissez-faire attitudes suffice during the transition period?
55:17Or are special social programs needed?
55:20Should available jobs be spread through overall shorter working hours?
55:25Must industry or government undertake substantial retraining and relocation programs?
55:31These questions have yet to be sufficiently studied.
55:35They hang over the future of American manufacturing.
55:39Over the ways in which the benefits of computer automation might be attained while mitigating its costs.
55:48Over the need to revitalize American industry and keep Americans working.
56:31For a transcript of this program, send $4.00 to NOVA, Box 322, Boston, Massachusetts, 02134.
56:38Please be sure to include the show title.
56:41This program was produced by WGBH-Boston.
56:45For a transcript of this program, send $4.00 to NOVA, Box 322, Boston, Massachusetts, 02134.
56:51Please be sure to include the show title.
56:54This program was produced by WGBH-Boston.
56:58This program was produced by WGBH-Boston, which is solely responsible for its content.
57:08Major funding for NOVA is provided by this station and other public television stations nationwide.
57:15Additional funding was provided by Allied Corporation,
57:18a world leader in advanced technology products for the aerospace, automotive, chemicals, and electronics industries.
57:26And the Johnson & Johnson family of companies, supplying health care products worldwide.
57:35To purchase film or video copies of this program for educational use, call toll-free 1-800-621-2131.
57:42In Illinois or Alaska, call Collect 312-940-1260.
57:55For a transcript of this program, send $4.00 to NOVA, Box 322, Boston, Massachusetts, 02134.