OUR HEARTLESS FRIENDS THE ROBOTS (May, 1963)
Excellent article and pretty accurate too. I loved that they made the early robots pay dues to the machinists union!
OUR HEARTLESS FRIENDS THE ROBOTS
By D. S. HALACY, JR.
WHEN a clock manufacturer needed production line workers recently for a ticklish assembly job, he ordered them from a firm called U.S.I. Robodyne. The workers weighed a bit over 50 pounds, and the clockmaker didn’t hire them—he bought them outright for about $2500.00. Slavery Involving midgets? No, these workers, each doing a man’s or woman’s job, are robots produced by the Robodyne Division of U. S. industries, Inc., at Silver Springs, Md. These “TransfeRobot 200” mechanical midgets, while not the first automated devices to displace human workers, are unique in some respects. First, they are not custom made, but are standard “off-the-shelf” items available immediately. Second, they are not one-job workers, but can be programmed to handle many production jobs within the scope of their electronic brains and mechanical fingers. Finally, they pay union dues!
Perhaps because they operate in a more dramatically human fashion than most automation equipment, TransfeRobots are in the spotlight of attention being given by both management and labor to the technological and economic problems created by progress. In addition to the clockwork assemblers already mentioned, there are many more such small robots building a variety of items —including typewriters and auto parts.
President John Snyder, Jr., of U.S. Industries, and Al Hayes, President of the International Association of Machinists, head a foundation which is working toward a smooth and painless-as-possible integration of automation into production work in this country. And that’s where the dues come in. Each TransfeRobot (via its employer) pays $25 a year as soon as it goes to work. U.S.I.’s larger equipment pays more, ranging up to a maximum of $1000 a year.
Critics, perhaps with human workers in mind, have described these machines as robot dolls—”you wind them up and they make money for the boss”—but both sides of the bargaining table realize that human workers must make money, too, or they can’t buy the goods the robots produce.
TransfeRobot has a big cousin called “Unimate.” Built by Unimation, Inc., of Bethel, Conn., Unimate costs ten times the modest price of the smaller robot. It weighs considerably more—a ton and a half—and it can heft loads of 75 pounds and exert a squeeze of 300 pounds with its steel fingers. Its builders list a hundred jobs that Unimate can do, including loading operations, assembly work, painting, welding, and similar tasks. It also has a brain, and can memorize 200 sequential movements after being “led by hand” through a new job just one time. Such a rapid learning capacity makes it sharper than the average worker, and Unimate is capable of round-the-clock operation without tiring, needs no coffee breaks, and is not distracted by pretty girls.
When Is a Robot a Robot? Since most of us have a rather vague knowledge of robots acquired by reading science fiction or watching the movies, it will be helpful to define just what is meant by the word. Webster calls a robot a mechanically efficient worker devoid of sensibility. Robots have other names, including “mechanical men” and “automata,” depending on who is doing the name-calling. The more sophisticated term goes well with automation.
Having defined the robot as a mechanical man of sorts, we realize that there are several narrower classifications possible within that general description. An automatic lathe, for example, is a machine capable of working by itself. So is a wrist watch. Less obvious, perhaps, is the time switch that turns on the furnace in the morning or the photocell system that turns on a light at dusk. Such devices rank fairly low on the robot scale.
The next step up the ladder is what some call a “proper” robot—a robot device which does not always function in exactly the same way. A more versatile fellow, the proper robot can cope with unpredictable changes in his environment. If we add a thermostat to our furnace control, or a switch to the corner traffic light so that it changes when a car rolls over it, we have a proper robot. The robot pilot in ships and aircraft is a highly developed proper robot.
There is another type of robot, the “true” robot, whose performance parallels that of an idealized human. The true robot is thus far fictional, but some scientists believe that the existence of man is proof enough that such a machine can be made. Less scientific minds jump to the romantic conclusion that this robot will even be man- or woman-shaped. Developments seem to bear out the former belief, at least, and we may one day be dealing with some very human-like robots; robots that are mobile, that listen and learn, think, show initiative, and act.
“Mobot,” “RUM” and “Beetle.” TransfeRobot and Unimate are still in the class of robots that simply do their jobs over and over. For factory work, of course, this is the best kind. A cousin of this simple plodding type is a robot that acts more flexibly; not with its own electronic brain but under the guidance of a human being. Impressive mechanical men of this ilk include Hughes Aircraft’s “Mobot” (for mobile robot).
An extension of the mind and hands of a human operator, such a robot works in high-radiation environments in nuclear plants, handles dangerous liquids, twists heavy iron bars, picks up eggs gently, and does even more ticklish tasks—such as fastening zippers for attractive young ladies, a chore that rattles some humans.
In 1960 Scripps Institution of Oceanography built “RUM” for the Navy—a Remote Underwater Manipulator which operated at depths of four miles. More recently, Shell Oil Company has used a Hughes Mobot in undersea oil explorations. And robot helicopters have been built, adding wings to the arms and legs of the mechanical man. But the majority of “mobots” developed so far are land-based. One of the newest, and surely the largest, is “Beetle.” Constructed by General Electric for the Air Force, this giant is used around missiles and was designed particularly for those fueled by nuclear devices.
Deep Into Space. The space age came on the heels of automation, and it is beginning to enlist the services of the robots. Plans to explore the moon include lunar “rovers” that will plod or roll or wiggle, depending on the type of surface they find on that satellite. NASA’s “Surveyor” is typical of such space robots, and it will busily poke around and report its findings to earth.
Deep space probes have no tin can man sitting at the controls, of course, but they are robot-manned, nonetheless. These robots read instruments, scan the skies for stars and planets and radiation, and act accordingly.
An interesting idea is that of a human pilot operating a spaceship by remote control using television for his eyes. Already in existence are TV receivers that fit the user like a helmet. The operator simply turns his head when he wants to look about, and the transmitter in the robot craft turns similarly. The sensation is described as being so realistic that the operator feels that he is in the distant craft. This idea of “tele-coupling” a man and machine seems to have an important future.
Robots That Think. Fascinating as these “mobots” are, other robots are far more intriguing. Operating a machine at the end of a wire, or even by remote control, is no very breath-taking concept despite the technical problems. And the precocious TransfeRobot is just a highly advanced wind-up man. More provocative is the idea of control of robots by the robots themselves.
Such an idea is not new. When James Watt put the flyball governor on his steam engine, he gave us the feedback principle that is the basis for automatic control. Thermostat-operated furnaces and float-controlled valves are simple examples of machine self-control. More recently, we have seen electronic computers exercising judgment in processing bank records and other paper work. Here, for all its size and unlikely appearance, we have a “proper” robot and perhaps the beginnings of a “true” one.
For all the pooh-poohing of the electronic brain, there are such devices as “Perceptron” that truly perceive. This electronic robot sees with photoelectric cells, learns to recognize things, and commits them to memory. There is another machine called “Artron” (for artificial neuron) that learns by reward and punishment in a fashion analogous to human learning. Still another robot, called “Cybertron,” solves “alogical” problems —those for which there is no formal answer and which require solution by trial and error. Prodded by a “goof button,” Cybertron handles tasks as varied as the classification of radar signals, and the grading of produce.
“Madaline” and “Hand.” Late in 1962, scientists at Stanford University demonstrated “Madaline I,” an advanced electronic robot that sees, hears, and feels. “Madaline” stands for Magnetic Adaptive Linear Neuron, and the demonstration included such feminine tasks as balancing a “broom” and taking dictation from the boss. Madaline has a mind of her own, made up of “memistors”— electrochemical resistors similar in function to human neurons. The word “adaptive” is the key to Madaline’s importance, for here is a robot not tied to a rigid program.
Many nervous watchers of developments have been happy with the fact that the robot brain and muscle have been kept safely separate, but the inevitable is beginning to happen. A young scientist at M.I.T. recently coupled an electronic computer with a mechanical hand-arm of the “mobot” type and created something he called simply “Hand.”
Thus far Hand is still in its babyhood and playing with blocks. In action, it carefully searches the surface of a table for such items. When it finds them, it picks them up and stacks them. It feels its way around obstacles, and when it finds an empty box, it explores the inside like a youngster delving into a cookie jar. If the box is the right size, Hand will store the blocks inside.
While Hand is visually blind, there are many robots that are not. Optical readers abound, and now there are machines that hear quite well, too. The Japanese have invented a typewriter that they call the “Sonotype”; it’s the lazy man’s dream–you just talk into it! In the U. S. there are computers like “Shoe-box,” so-called because of its size; unlike a real shoebox, it accepts verbal questions and gives verbal answers. Robots, then, not only think and act, but see, hear, and talk.
Robot Baby-Sitters? Years back, robots were suggested as companions for children: combination baby-sitters, tutors, confidantes, and all-around good chums. More recently the idea has been extended to the robot as a handy helper around the home. He would answer the phone and take messages, help with the budget and other problems, remind us of our appointments, and so on.
Only the child’s companion idea has been implemented so far, and this on a far more childish scale than proponents of the notion had in mind. Toy manufacturers have come up with a variety of walking, talking, command-obeying robots that are mighty popular at Christmas time. Shaped in the best science fiction movie tradition, with halting awkward stride and impressively blinking lights, these junior robots have one big flaw in that they cannot defend themselves. The death rate is terrific.
Robot Animals. When we leave the world of mechanical men for mechanical animals, we find some very impressive robots. Brain expert Dr. W. Grey Walter of England’s Burden Neurological Institute has built a number of electromechanical beasts physically resembling turtles. Dr. Walter prefers names like machina speculatrix, for their apparent ability to speculate.
Although equipped with only two “brain cells,” the first of these animal robots was capable of several responses to outside stimuli. Using its sight and touch organs, it circled curiously about a room, backing away from obstacles and shunning uneven surfaces. Seeing itself in the mirror, machina speculatrix almost seemed to preen. When it got hungry (because of waning storage batteries), the robot turtle sought out its den to feed on an electrical outlet!
Walter created a more intelligent machina docilis that could learn, and “CORA,” for Conditioned Reflex Analogue. CORA, like Pavlov’s dog, learned from hearing a whistle and from being kicked. She also exhibited frustration in the face of conflicting orders, a creditable performance for a six-celled brain.
Fellow Britisher W. Ross Ashby built a homeostatic robot which demonstrated, among other remarkable qualities, that of “ultra-stability.” A conventional aircraft robot pilot is connected to the controls in such a way that displacement of the plane from normal will bring about a proper righting force. If the controls were hooked up backward, however, the robot would blindly fly the plane to disaster. Not so the ultra-stable robot, or homeostat. It will seek a stable position no matter how it is wired, much as man adapts to a radically changing environment.
An interesting robot animal was built by communications expert Claude Shannon. On a visit to England he blundered his way through a famous hedge maze in about 20 minutes and got to thinking about such a problem in relation to telephone switching circuits, his own province. Shannon labored and brought forth a mouse. This was a very special mouse, however, and it could run a maze in remarkable fashion. The maze consisted of 25 squares with removable partitions that made possible a million different routes. Placed on any square, the robot mouse could find his way to the cheese in about two minutes of trial and error bumping. On the second run it followed an errorless direct route in the fantastic time of 15 seconds! This is a feat far superior to that performed by any real mouse—or man!
Shannon’s mouse was named “Theseus” for the ancient Greek who successfully negotiated another maze in another time. No robot, Theseus was human enough to require a ball of yarn to find his way through the labyrinth. But the idea of robots is as old or older than Theseus. The Iliad describes golden, three-wheeled mechanisms that served as information carriers for the God Haephaestus, and the Old Testament tells of “golems” who were early-day robots run amuck.
It is often difficult to tell which came first, fact or fiction, and real mechanical men have almost as long a history as the stories about them. Eli Whitney and his plaintive cry of “Keep your cotton-picking hands off my gin!” were contemporary with the doomed Dr. Frankenstein; and the year the play R. U. R. introduced the word “robot” to the world, the first automatic factory for turning out chassis for cars went into operation in the United States.
The scratchings of the machines on the wall were so obvious by 1946 that an article in Fortune contained the disquieting news that “the human machine-tender is at best a makeshift.” Two important developments were described as part and parcel of the new kind of factory. One was the electronic computer for monitoring and controlling operations; the other was the robot “hand-arm” to implement these orders.
Age of the Robot. Robots, then, have not burst full-blown upon the current scene but have a long, seesawing history in which science and fiction have tried to outdo each other. We are, however, entering the important phase of “robotry”—a phase which has had to wait for a number of factors to be right. Among these are economic need, maturing of concepts and technology, and popular acceptance. Where historically the robot has been employed as mechanical bogeyman and stuntman, we are now seeing him gainfully employed.
Man’s inherent laziness caused him to create the robot; his guilty conscience makes him fear it. However, where once we worried about machines going wild with destructive results, and searched our souls to justify this tampering in the domain of the Almighty, most of the fear today is more realistic. While few advocate stoning the machines and killing their builders, most recognize that this phase of the industrial revolution is not without its painful upsets.
Granted that we need automation and its computers and robots, and that the alternative is to “give us all pointed sticks and have us go plant rice in the paddies,” the technological unemployment being discussed is no union-inspired bugaboo. Integration is the topic today, and perhaps we should include the integration of the machine into society.
One ghost should be laid to rest, however: the fear that thinking robots will make our own brains wither away. Years ago many predicted such a fate for our muscles when mechanical transportation became widespread. These doomsayers forgot the old-time cowboy who wouldn’t walk across the street if his horse was within a mile; they couldn’t know that the first four-minute mile, seven-foot high jump, and fifteen-foot pole vault would come long after man would supposedly have atrophied into two hands to grip a steering wheel and a right foot to push the gas pedal. For the same reasons, our brains are not going to shrivel either. Electronic computers have already freed scientists of much drudgery so that they can spend more time on true creativity. Thus, our brains are actually more productive.
As the number of robots grows, and they even learn to reproduce themselves, the question is no longer whether or not they are going to take over. It is simply how we are going to get along with them now that they are doing it. Assembly line worker TransfeRobot 200 is a case in point. As mentioned earlier, the TransfeRobot’s annual “dues” are being used to finance intelligent studies of the problem in a foundation set up by U. S. Industries, Inc., and the International Association of Machinists. Such studies, we hope, will show that dictionary definitions to the contrary, the robot has a heart after all.