Thin clouds of smoke and steam drift through the huge dark room, as the heavy steel dies of the die-casting machines slam shut, then explode open. Mechanically, an arm reaches into a machine, grabs a piece that's been cast, lifts it out, then, turning, hange it on a rack. At The Newton-New Haven Co., a $12-million-a-year die-casting plant in North Haven, Conn., the industrial robot is already earning its keep.

"They give us a distinct competitive edge," says Newton-New Haven's president, James R. McClurken, referring to his seven Unimate 2000 robots. Once considered the exclusive property of giant automobile manufacturers like General Motors and Ford, the mechanical workers are turning up with increasing frequency in small companies. Like McClurken, the owners of these firms find that using robots makes good economic sense. In a typical die-cast operation, for instance, the introduction of a robot generally leads to a 30% increase in productivity; other industries have reported increases of up to 300%.

McClurken got hooked on the technology more than 20 years ago, when he was plant manager for one of the largest die-casting operations in the country, Pottstown, Pa.-based Doehler-Jarvis. Die casting, which is used to produce an incredible variety of metal parts for other manufacturers -- everything from components for meat slicers to billiard table equipment -- requires workers to handle metals reaching 1,250 degrees F., at high pressures on machines that may knock out hundreds of units per day. Under these conditions, heavy turnover was a constant problem at Doehler-Jarvis. "A diecasting plant is a real rough environment for a man to work in all day," says McClurken. "We were constantly training new operators. We felt dammit, there's got to be a better way."

That was in 1961, the same year that McClurken met Joseph F. Engelberger, of Danbury, Conn.-based Unimation Inc., one of the first robot manufacturers in the country. None of his competitors had yet sold a robot to a die-casting plant, and Engelberger, anxious to get a foothold in the industry, approached McClurken at Doehler-Jarvis. "We didn't buy the robot," says McClurken. "We leased it. Unimation wanted to get some installed, and it was such a good deal that I couldn't turn it down." Machine operators nicknamed the robot "Knucklehead."

Once Knucklehead was installed -- floor layouts had to be changed and new water lines put in -- and its bugs worked out, the robot performed like a, well, like a robot. The machine had a number of advantages, besides providing labor for an increasingly unpopular job. Knucklehead worked three shifts without coffee, lunch, or bathroom breaks, broke down less frequently than the diecasting machines, which dated from the early 1900s, and performed more consistently than a human worker. Altogether, it produced a better percentage of good casts and reduced machine maintenance. "There were unexpected cost savings that were very significant," says McClurken.

Doehler-Jarvis's strong UAW local didn't object to the robot, in part because both the company and Unimation were careful not to displace workers by the new equipment. "We added robots as attrition took care of the people doing that job," says McClurken. Notes Unimation's Engelberger, "Our policy is that no one can ever point to a Unimate and say, 'That machine took my job." By 1973, when McClurken left Pottstown, Unimates were tending 25 of the plant's 60 die-casting machines.

Not surprisingly, when he joined Newton-New Haven four years ago as executive vice-president, one of McClurken's first recommendations was that the privately held company go with the new technology. The 60-year-old firm had been hard hit by the 1975 recession, experiencing a nonprofit year for the first time in 22 years, and William Newton, Jr., then the firm's president, had been forced to lay off 70 of his 180 employees. Writing about the experience a few years later ("Recession: Do You Have a Contingency Plan?" INC., September 1979), Newton had admitted, "Like a lot of small companies, we're sometimes guilty of not seeing what's happening right under our noses." This time, though, Newton was determined not to make that mistake -- Newton-New Haven bought its first robot.

McClurken, who had gained considerable experience with die-casting robots by that time, oversaw a smooth transition. "With this machine," says McClurken, "once you bring it in and set it on the floor in the right place and plug it in, if you know what you want, you can get it going in a week." Other units were added about once every three months, and again, the phasing in was done within attrition levels.

Seven of the company's die-casting machines are now operated by Unimate 2000s, a 3,500-1b., general-purpose robot with a base price of $46,000. All of Newton-New Haven's units are factory-rebuilt models. McClurken would like to add more, but is holding off; his plant's layout is antiquated, and some of the older machines can't be easily adapted to the technology. He is, however, contemplating the installation of some of Unimation's smaller Puma robots in the metal-working area. These will perform secondary operations like trimming, drilling, threading, and assembling cast parts.

The Unimate 2000 is a tank-shaped unit with a rectangular base supporting a turret-like device from which a steel arm extends. At Newton-New Haven it's used to grab a so-called gate (a rough casting) from the die, bring it out, and simultaneously spray lubricant on the die faces for the next cycle. As the gate comes out, the robot draws it past a sensor, which tells the robot whether it has a complete casting or whether part of it fell off or stuck in the die. If the casting is complete, the robot trips a mechanism to close the machine. If the casting is incomplete, the Unimate shuts the machine down and calls for help. With a good casting, the robot then swings around, hangs it up, and reaches for the next. It's a virtually foolproof system, and one that has paid off for Newton-New Haven. "If we took out the Unimates," says McClurken, "we'd probably have to hire at least 12 to 15 more people."

Each of those operators, he estimates, would earn about $7.50 an hour; with benefits included, they'd wind up costing more than $10 an hour. A Unimate 2000 -- even when all costs of acquiring and operating it are factored in -- works for less than $6 an hour, which comes to an annual savings of $156,000 -- $216,000 in operator costs alone. When reductions in scrap and maintenance are added in, the savings mount even higher. And a Unimate never quits or dies. The units can be rebuilt by the manufacturer virtually ad infinitum.

"You can also run more hours per day on a consistent basis with a Unimate than you can with manual operation," says McClurken. "Since we're really selling a manufacturing service -- our talent and the time on our machines -- hours of operations is crucial."

McClurken estimates that the robots trim his breakeven point by about $150,000. With larger orders (short runs don't justify the use of a Unimate) or improved plant layout, the figure could be more significant. "They're capable of greater utilization than we've been able to attain so far," he says. As far as McClurken is concerned, the more "Knuckleheads" there are in the world, the better.

Webster's New Collegiate Dictionary offers a more precise description of a robot than a "Knucklehead." It defines a robot as "a machine that looks like a human being and performs various complex acts (as walking or talking) of a human being" -- like R2D2 or C3P0 or some similar Star Wars contraption. But the Robot Institute of America, the Dearborn, Mich.-based trade association for robot manufacturers, distributors, and users, defines the robot much more narrowly: It is, according to the institute, "a reprogrammable multifunctional manipulator designed to move material, parts, tools, or specialized devices, through variable programmed motions to accomplish a variety of tasks." Rather than cute chrome-plated midgets, the most common units are simply elaborate mechanical arms. Fitted with "fingers" -- everything from grippers to wrenches to spray guns to arc-welding equipment -- and programmed either by mechanical means or by a computer, the arm becomes a robot.

And what does the industrial robot do? Quite simply, it moves something from one point to another.

But it does so with amazing accuracy and dependability.

Today's robots are used principally for spot welding, spray painting, die casting, investment casting (making molds in sand), loading and unloading machine tools, stacking items on pallets, and plastic molding. They're also used in foundry, press, metal deburring, and brick and glass handling work. Experts predict robots will have a major impact on arc-welding and assembly operations; their use in the latter application is expected to quadruple by 1990.

In a factory, robots may position a spot-welding unit on the seams of an automobile's shell, float a spray gun across a refrigerator, or pack bottles of cologne in gift-wrapped containers.

In a Chesebrough-Pond's plant in Huntsville, Ala., a Puma 500 robot outfitted with a miniature squirt gun sprays liquid cosmetics, such as eye shadow and cheek blush, into tiny metal pans that will be inserted into Prince Matchabelli compacts. At another of the company's plants, one in Clinton, Conn., a robot places glass bottles onto an assembly line where they will be filled with one of the components of Rave home permanent; the robot plunks down 8 jars at a time, about 130 jars each minute.

Robots handle plutonium for the U.S. Department of Energy, make light-bulb filaments for Westinghouse, drill parts for the F-16 fighter for General Dynamics, assemble motorcycle engines for Yamaha, and, at one robot factory in Japan, manufacture more robots. Futurists envision robots mining the deep-seabed, serving drinks to partying executives, and waging war in place of the common foot soldier.

For now, however, they've been relegated to simpler moving tasks. Speed, accuracy, payload size, the type of power they use, and memory capacity are among the criteria that determine how well suited a robot is for a given task. Robots are also rated by axes of motion (essentially the number of directions in which the arm and its "fingers" can move -- three to six is standard), repeatability (how accurately it will return to a given point in space during normal operation), and velocity. A Unimate 760, for instance, can maneuver a 22-1b. load in six directions at a speed of 1,000 mm/sec., bringing it to within.008 inches of a predetermined point, over and over again.

Robots may be driven pneumatically, hydraulically, or electrically, and are programmed either by mechanical means, such as a step drum (not unlike a player-piano roll), or by a computer (increasingly, the brain of choice).

Two-dimension vision systems which employ TV cameras are already being used to instruct robots -- to tell them the location, for instance, of a particular part on an assembly line -- and further refinements are in the works. Future systems will be better able to gauge all three dimensions, one day making it possible for a robot to spot an obstacle and maneuver around it, or to pluck different pieces from a container and assemble them as easily as a human could. Engineers are working to make robot arms lighter and more flexible, but most of the current research is aimed at improving the robots' ability to see, hear, feel, and possibly smell. Within the near future, they will respond capably to voice commands and will "feel" so accurately that they will be able to handle, if not juggle, fresh eggs. Other pending developments include multiple arms, hand-to-hand coordination, mobility, and self-diagnostic capabilities.

The modern robot is based in large part on patents applied for in the 1950s by George C. Devol, a self-taught inventor who, at the age of 70, now heads Devol Research Associates, a robotics leasing and consulting firm in Ft. Lauderdale, Fla. In 1961, Unimation installed its first Unimate in a General Motors plant, and AMF Inc. introduced its Versatran into a Ford facility. Since that time, more than 6,000 robots have been sold in the United States, most of them to the automobile and appliance industries. Of that number, approximately 4,100 remain in use today.

Currently, there are about 30 U.S. robot manufacturers, as well as robot vendors, robot systems companies, robot consultants, robot analysts, robot industry suppliers, and robot rental firms. Unimation is the largest manufacturer, with a 39% share of the market; other leaders are Cincinnati Milacron Inc., DeVilbiss Co., Prab Robots Inc., and ASEA Inc. Recently a number of large companies, including IBM, Bendix, General Electric, and Westinghouse, have entered the field. Sales by U.S. manufacturers have climbed from $35.5 million in 1978 to $155.5 million in 1981; this year, they are expected to hit $215 million, and, by 1990, should be in the $1-billion-to-$2-billion range.

At the Robots VI Conference and Exposition held in Detroit in March 1982, about 100 exhibitors demonstrated their products for nearly 28,000 businesspeople, bankers, engineers, and students. At one point the crowds grew so large that the fire marshal was forced to shut down the exposition; visitors offered to buy robots off the floor, and, at the end of the four-day affair, one salesperson was overheard to say, "It sure feels nice to be going home with 52 orders." Clearly, robots are generating a great deal of interest.

The reasons are numerous, but they all go to the bottom line: "The primary reason for using robots is economics," says Unimation's Engelberger.The savings and advantages accrue at virtually every step.Conventional machines, if they exist to perform a particular job, may cost three to five times as much as a robot; one aerospace firm, for example, is using an $85,000 robot for work that would otherwise require a $300,000 machine. And once installed, conventional machines resist change. In the past, automobile manufacturers had to shut down for months in order to retool assembly lines; now, because robots are easily reprogrammed, the same result can sometimes be achieved in days. Conventional machines may also require greater maintenance.

The costs of running a robot are equally attractive. "If you buy a $50,000 robot," says Engelberger, "pay for the loan, install the robot, depreciate it, maintain it, overhaul it, and give it an eight-year life, the total cost is about $6 an hour." The ordinary automobile worker, by contrast, now costs an employer about $19 an hour. "General Motors says that it's going to have 14,000 robots by 1990," Engelberger points out. "If they could do it just one year sooner, it would mean $728 million per year in savings."

The savings, generally calculated on the basis of a two-shift day, are less dramatic but equally important in the case of a small business. The rebuilt Unimates used at Newton-New Haven cost $4 less an hour than human operators. If the robots are run for three shifts, the savings go even higher. Payback on a purchased unit ranges from one to two and a half years, but robots don't have to be purchased -- they can be rented or leased.

"For $1,000 a month," says Walt Weisel, vice-president of Prab Robots Inc., in Kalamazoo, Mich., "you can lease a robot for three years and it will work for you shift after shift."

There are other savings that, though less easily calculated, demand consideration. Among these are reduced downtime (the average robot is "up" 98% of the time), less scrap, a higher quality product, and improved safety. (A robot also uses less energy than conventional machining, and doesn't need warmth, clean air, or even light.) The only thing wrong with robots, as far as many people are concerned, is that they are an underutilized resource.

"There's little in manufacturing that can't be done by robots," says Weisel, noting that Prab has placed units with Eastman Kodak, GE, Westinghouse, IBM, Deere & Co., International Harvester, GTE, and a host of smaller companies. The quality of robots continues to improve and, as more manufacturers have appeared, their cost has dropped. Prab recently chopped $10,000 from the price of some of its models. Why then have such a relatively small number of companies, particularly smaller ones, made use of the technology?

Even Engelberger admits that robots operate within certain constraints. He notes that they work best at repetitive tasks and are only cost-effective if these jobs would ordinarily be performed by high-priced labor; they require an organized work space (older plants may find it uneconomical to revamp their layout); they need an inventory of replacement parts and a person who knows how to use them; there are some jobs they are not yet "smart" enough to do. But the sociological hurdles seems to have played a larger role.

"All of manufacturing is extremely conservative about new technology," Engelberger explains. "One of the blows to me in my career was to find out that no one needs a robot -- that's a shocking discovery when you think it's such a wonderful product.

"What we had to do was find a wedge [into the marketplace], and what's happening today is that our wedge is coming from Japan. Now it's a different story; now manufacturers better have robots or there won't be manufacturers anymore. The United States will wind up being a service industry."

Gene Macri, vice-president of Productivity Systems Inc., a Farmington, Mich.-based robotics consulting firm, notes that the greatest resistance to robots comes from middle management. Middle managers, he says, "avoid risk by doing things the way they have always been done." They fear the temporary downtime that installing a robot may mean, even though the technology could increase profits in the long run. If things go wrong, the managers say, it will be their necks, not those of the robot manufacturers, that are on the line. "And old-timers," Macri adds, "may also fear that the new equipment could strip them of their expertise, which is predicated on old systems and old equipment."

Even robot makers seem to have lacked vision at the outset. It was General Motors that "came up with the idea of doing spot welding," Engelberger concedes. "We never thought of that." (Robots for spot welding on automotive lines now account for 57% of Unimation's business). "There's a great deal of serendipity involved in this robotics game."

If manufacturers had been a bit more imaginative in their early marketing, Engelberger says, robots might now be used in a broad range of businesses rather than concentrated in less than a dozen heavy industries. And, once manufacturers latched onto big companies, they neglected the smaller ones; as a result, 20% of the clients now buy 80% of the product. But that's beginning to change.

Laura Conigliaro, an analyst with Bache Halsey Stuart Shields Inc., points out that the recent recession, which played havoc with the auto and appliance industries, is forcing some major rethinking on the part of robot vendors. "Ultimately," she explains, "the effect of this downturn will be positive, as robot vendors realize that other markets must be pursued vigorously."

Surprisingly, organized labor has offered little resistance to the new technology. Employees have generally been replaced during normal attrition, and displaced workers are frequently upgraded into better jobs. Labor leaders fear that industries -- not jobs -- may be lost to Japan.

"We don't have 300,000 auto workers out of work because we put in robots," says Weisel. "It's because the Japanese put in robots and are bringing their products here."

Workers generally accept the automated assistants after some initial trepidation. One welder, whose partner had been "bumped" by a robot, said of it, "I like it fine; it doesn't talk back to me." And when a robot at a Ford stamping plant in Chicago broke down for an extended period, its human co-workers sent it cards and threw it a get-well party.

Businesspeople are becoming more aware of he advantages of using robots. More and more are hiring robots, and many of these "Knuckleheads" are going to work for small companies. For years, there was a truism that indicated, "If you need only one robot, you don't need any." Given more reliable, less expensive, and easier-to-use robots, escalating manufacturing costs, stagnant productivity, and an economy that seems determined to sink the small businessperson, that is no longer the case.

"We've got robots in die-casting shops that had five people at the time we put them in," says Weisel. These five included "the owner, his wife, who was keeping the books, his son, who swept up afterwards, and two workers."

Robots are already earning their keep at the Shelby Die Casting Co. Inc. in its Fayette, Ala., plant; Engineered Sinterings and Plastics Inc. in Watertown, Conn.; Kerkau Manufacturing in Bay City, Mich.; and at hundreds of other small firms throughout the United States. Kerkau, a small custom and production machine shop (37 employees, annual sales of about $2.4 million), discovered that it was able to run a single robot on a part-time basis and still come out ahead.

Five years ago the company, which specializes in the production of valve components and related piping parts for the oil and chemical industries, installed a new machine and a Prab robot to run it.The Bullard machine, which turns, faces, bores, and threads pipe flanges, is used only when the company gets an order for at least 3,000 units; the Prab loads the raw flanges onto the machine, turns them over halfway through the eight-step process, and then unloads the finished product.

Don Griffin, who oversees Kerkau's machines, estimates that the Prab is "at least 50% faster than a man." Adds sales manager Mick Weadock, "With some of the larger flanges, if you had a guy doing it, his arm would probably fall off. You get pretty darned tired loading and unloading 35-pound pieces."

The robot, he notes, has been "very cost-effective" and has been less trouble to maintain than many of the plant's other machines. "Everyone's pleased with it," he says.

At the Robots VI Conference and Exposition in Detroit, robot manufacturers, users, and cognoscenti seemed to have reached a consensus, stated most succinctly by Engelberger, who believes that the American manufacturer now faces a critical choice. "Automate, emigrate... or evaporate," He advises.