Case study: Unimation, Inc., genius, innovation, inspired invention and rapid decline.
Fifty years ago this month (October 8, 1962) Joesph Engelberger strolled into the Secretary of State’s office in Connecticut and filled out papers for his new corporation, Unimation, Inc., the world’s first industrial robot manufacturer.
That same year, angel investor Pullman Inc., developer of the famous railroad car of the same name, put up $3M for a 51% share in Unimation, Inc. ($1M in 1962 equals $7.6M in today’s dollars), a company that wouldn’t turn a profit for fifteen years.
Lots can be learned from the strange tale of Unimation Inc., and how it went awry. High expectations, elation, a fatal flaw, and finally the sobering reality that Unimation would be no more.
There from the beginning
George Munson, there from the beginning, recounts the rise and fall of the world’s first industrial robot builder.
GEORGE MUNSON, “PITY THE PIONEER: THE RISE AND FALL OF UNIMATION, INC:” In the spring of 1951 the Korean War was in full swing, and I was sure I would be drafted. I saw no point in interviewing for employment, despite my newly awarded degree in physics from the University of Connecticut.
Meeting Joe Engelberger
When I heard about a starting position for a physicist at Manning, Maxwell and Moore (MM&M) in Bridgeport, Conn., I figured I had nothing to lose and made an appointment. A young engineer, sporting a bowtie, by the name of Joseph Engelberger, interviewed me – he hired me on the spot. Little did I suspect that this decision sealed my fate, as our association would change manufacturing the world over. Nor could I know that with his combination of entrepreneurship, marketing, and natural affinity for promotion he would become the “Father of Robotics.”
MM&M specialized in a variety of electronic devices. Joe had established their Aircraft Products Division. Military spending was up and we landed some lucrative subcontracts for jet engine controls. With brisk business that included the U.S. Air Force, we outgrew the company headquarters in Stratford, Conn. and in 1954 moved into our own plant in Danbury, Conn.
The war wound down and so did our business. MM&M ordered Joe to liquidate the division, but instead, he began looking for a stable line of work to keep his workforce together. He bought five books on finance, sat down and read them all. In his words, “I got my MBA in one weekend.”
Programmed Article Transfer a/k/a robot
In 1957, Joe met the creative genius George Devol. Born in Louisville, Kentucky in 1912, Devol circumvented formal education and, at age 20, formed United Cinephone, producing recording equipment using photocells. He developed the revolutionary barcode and patented hundreds of inventions, including digital magnetic recording.
Devol had observed mountains of scrap tooling, created by product design changes. It inspired his revolutionary idea – universal automation; automation that would not become obsolete, but would adapt to product changes. He also patented a device that could perform repeated tasks with greater precision and endurance than the human worker, at less cost, and be retrainable for new tasks. He called it Programmed Article Transfer, and later “Robot.”
Devol facilitated the sale of the Aircraft Products Division to Consolidated Electric Corp. (later Condec) and I stayed on with their new company in Bethel, Conn. – Consolidated Control Corp.
Engelberger the entrepreneur and Devol the inventor collaborate.
Soon, Joe convinced Condec’s CEO, Norman Schafler to finance Devol’s brainchild, the industrial robot. If Joseph Engelberger was the Father of Robotics, George Devol was the grandfather. Interestingly, the word robotics was coined by one of Joe’s fellow Columbia University alumni, Isaac Asimov. Joe received his master’s in physics in 1949 and Asimov received his Ph.D. in chemistry in 1948. Joe found Asimov’s books about robots inspiring. In his forward to Engelberger’s book, “Robots In Practice,” Asimov includes his Three Laws of Robotics:
- A robot may not injure a human being or, through inaction, allow a human being to come to harm.
- A robot must obey the orders given to it by human beings, except where such orders would conflict with the First Law.
- A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws.
How we began
Our beginnings were modest, only six of us were assigned to the project. We knew the robot had to be anthropomorphic, but which configuration would provide the greatest flexibility for the applications we foresaw?
First, we conducted market surveys to determine the parameters, considering four basic configurations: polar, cylindrical, Cartesian, and revolute coordinates. Our decision was to proceed with a polar coordinate design, and while a 6-axis machine would have best emulated the flexibility of the human arm and wrist, the expense and complication forced us to build to a 5-degree of freedom machine, having just two rather than three wrist axes.
With the robot’s configuration determined we began to develop the prototype. A self-contained hydraulic supply operating at about 1,000 psi would provide sufficient power and require fewer gears, thus, less backlash, than an electric motor. However, hydraulic power technology was not advanced enough, and the demands for speed, stability and accuracy challenged every design aspect of the 2,700-pound behemoth. Our engineering design tasks included:
1. A digitally controlled system based on the binary system. (Remember, this was in 1956!)
2. A nonvolatile solid-state memory system, which didn’t yet exist.
3. Shaft position optical digital encoders for high-speed performance, which also didn’t exist.
4. A high-performance digital servo controller capable of dynamic control with a wide range of payloads.
5. High-performance hydraulic servo valves.
6. Self-contained electrical and hydraulic power supplies.
A ferroresonant sensor
Under Devol, we developed a ferroresonant sensor, the basis for a self-styled memory system, patented as “Dynastat.” We also needed an optical shaft position encoder to provide the necessary position feedback to close the loop between the robot arm’s actual position and its command positions. By 1965 we had perfected an optical Gray code encoder we called ‘Spirodisk.”
We put together a hydraulically driven programmable arm that could pick up metal letters and spell out short phrases, and in 1961 we introduced our robot at a trade show at Chicago’s Cow Palace. Nobody knew what we were displaying or why, and the hydraulic system leaked like a sieve, but we were on our way! We needed a company and product name. “Universal automation” contracted to Unimation and the industrial robot was born. We called it the Unimate.
Ford doesn’t get it, but the Japanese do – big time!
Engelberger and Devol now approached the Ford Motor Co. The Unimate got the attention of the VP of Production, who proclaimed that he could use “thousands of them.” A manufacturing engineer was assigned to “do something” with the specification, but he passed it on to suppliers who might be interested – FoMoCo did not have the vision to recognize the robot’s potential in assembling car bodies until much later.
WWII and the Korean War stimulated many new products and manufacturing technologies in the U.S., which led to a large dose of complacency. Thus, America’s successes gave way to international competition – notably from Japan – that was unforeseen and, eventually, unstoppable. Contrary to popular belief even now, robotics is not a Japanese-founded technology. It was exported from Versatran and Unimation in the U.S.
In the early 1960s, when we began our mission to revolutionize American manufacturing, labor was abundant and competition from abroad was not yet threatening. American manufacturing knew no real competition. It wasn’t until the 1970s that the rivalry of the Japanese awakened American industry to its vulnerability.
As the highest paid workers in the world, Americans were competing with workers in every other nation, who were paid far less. This was not a favorable atmosphere for our product, which many viewed as frivolous.
Japan was fighting the perception that “Made in Japan” meant shoddy goods. In response, they set out to produce quality goods, heeding W. Edwards Deming’s Total Quality Management philosophy: Quality carries with it reduced costs and improved competitiveness. Deming’s fellow Americans did not grasp his philosophy until much later, to their detriment. With the pressure on, American industrial leaders had to rethink their position.
While justification for the robot, both economically and socially, seemed obvious to us, we were not surprised to find resistance in the workforce, particularly in mass production industries – the social threat was seen as devastating. But economic justification was far from obvious for even the most forward thinking accountants. We had our work cut out for us.
Impact of robots on job displacement and job creation
Interestingly, a 1983 study of human resource implications of robotics concluded, “The most remarkable thing about job displacement and job creation impacts of robots is the skill-twist that emerges so clearly when the jobs eliminated are compared to the jobs created.
“The jobs eliminated are semiskilled or unskilled, while the jobs created require significant technical background. We submit this is the true meaning of the robotics revolution.” It appears that we were prophetic, as this study revealed what we had earnestly stated since the mid-60s.
Everything needs a killer app
In 1961 we got our opportunity to put our innovation to the test at GM’s die-casting plant in Trenton, NJ. In wild anticipation, we readied Serial Number 001 for shipment. Naturally, we were concerned about how the diecast machine operators would react to this man replacement. In fact, their consensus was that our machine was a curiosity destined to fail.
However, until the application of the robot to spot welding automobile bodies came along in the late 1960s, no other industry encouraged the proliferation of the industrial robot like die casting. It inherently required all of the attributes the robot had to offer. Eventually some 450 Unimate robots were employed in die casting.
In 1962, Pullman Inc., became a silent partner in Unimation, investing $3M to buy the high-tech element they desired for their corporate structure. The Pullman people, from the top down, were good people and it was a pleasure to be associated with them, but in the mid-70s new management lost interest. Condec bought Pullman’s 51% interest in the company and remained sole owner until Unimation went public in 1981.
In our efforts to broaden our customer base it was natural that we should strike out internationally. Since our cash flow was entirely negative, we needed a partner. We chose the largest manufacturing business in the world – Guest, Keen & Nettlefold in Wales. GKN was enthusiastic and committed. We offered strong support, educating their engineers in robot “lore” and guiding the sales force. But during my 1966 visit I observed that every plant had old machines and old methods.
Good applications were few and far between, so the relationship was dissolved. Joe then set up Unimation Ltd. in Telford, England, which led to considerable business throughout Europe and Scandinavia. Then, in 1966 he licensed Finland’s Nokia, Ab. to market robots in Scandinavia and Eastern Europe.
Intellectual property and Japan
While Unimation was establishing the robot’s credentials, Japan was enjoying economic prosperity, but anticipating a labor shortage. Thus, in 1967 Joe was invited to give a lecture in Tokyo to a large group of engineers, which included a marathon, 5-hour Q&A session. This interaction culminated in a 1969 licensing agreement with Kawasaki Heavy Industries to manufacture and market Unimate robots for the Asian market – a good marriage that endured and prospered for 15 years.
By 1983 they had shipped over 2,400 Japan-made Kawasaki Unimate robots.
Joe wrote to several manufacturers suggesting that they might want to look closely at their technology and our patents. As a result, ASEA, Cincinnati Milacron, and IBM became licensees from which we derived royalties on their robot sales.
As George Devol had learned long ago, ownership of patents is a valuable asset, from which we benefited handsomely. They protected our intellectual properties and helped us develop a strong licensing position that lasted years. It was a gratifying position, but they were eroding our market share and making our technology obsolete.
First triumph of the robots: revolutionizing auto making
The automotive industry was the engine that drove the American economy, so we concentrated our energies there. Until the end of the 1960s, the auto body assembly line was a moving conveyor on which major body subassemblies were hung.
Our nemesis was that it used a level of skill and intelligence the robot didn’t have. Yet we knew that the big payoff was in applying the robot to the body assembly lines. This required a product designed for both automation and indexed conveyors.
These measures would open up many other applications, yielding a superior product while reducing cost. We now needed a champion at a high corporate level with insight, foresight, and guts. GM’s plant manager Les Richards had all three.
GM had rebuilt its plant in Lordstown, Ohio in 1969, making it the most automated automotive plant in the world, building 110 cars per hour [utilizing 28 spotwelding robots), twice the rate of any plant then in existence. Lordstown was to be the answer to Japan’s onslaught. It was to produce a high-quality small car [ Vega, and later, the Impala] that would satisfy the American public at a competitive cost, putting GM back on top.
The technological impact of the Lordstown experiment revolutionized automobile making and secured the robot’s place. It wasn’t long before other companies turned to robotics and indexing systems for a more disciplined approach to manufacturing.
At the same time, the European market came alive with Unimates at Fiat, Volvo, Mercedes Benz, British Leyland, BMW. Their unions welcomed robots performing all of the dangerous jobs.;
The activity in the auto industry created many nonautomotive employment opportunities, as diverse industries sought to improve their position through the application of technology: Bendix, Pratt & Whitney, Dupont, Whirlpool, GE, and many others. For some time our only competitor was Cincinnati Milacron of Ohio. This changed radically in the late 1970s when Japanese conglomerates began producing industrial robots.
Success evolves into a family of robots
Our robot “family” grew with each new application’s demands. The original 1900 Series developed into other series with extended reach; increased repeatability and lift capacity; 6-degrees of freedom; stronger wrists; Univision. New technical innovations were incorporated: solid-state memory; microprocessors; high performance electric motors replaced hydraulic and pneumatic ones; transistor controls; new programming languages.
In 1977, Joe shrewdly bought Victor Scheinman’s company, Vicarm, and renamed it Unimation West. We further developed Scheinman’s robot into the Programmable Universal Machine for Assembly (PUMA). We also acquired his VAL language, which was cutting edge.
Unimation also attempted to augment its line of machines by entering into marketing license agreements with other manufacturers, such as Trallfa of Norway and Electrolux of Sweden, but all were only mildly successful. The bottom line was that we were not very good at marketing someone else’s product.
Fatal flaw: blind eye to new technology
What happened next marked the beginning of the end. Joe’s innate business sense failed him, and his rigidity skewed his judgment in a critical decision.
In 1981, the 9000 with VAL was Unimation’s answer to all the competitors who were seriously eroding its customer base.
But it could not overcome one great deficiency – hydraulics. The auto industry wanted electric-driven robots. Joe balked at this, convinced that the muscular robots required had to be hydraulically driven.
He met with GM’s CEO to argue his position, but lost. The partnership that was struck in 1961 virtually ended at that meeting. To rub salt in the wound, GM then announced its partnership with Fujitsu/Fanuc to market Fanuc’s line of electric robots.
By 1981 Unimation was carrying long-term debt of $19M, owed to Condec for development of the robot. Schafler, pressured to generate more cash, restructured the company. Paul Allegretto, a Condec executive, was made Executive VP and I was named VP and General Manager of the newly created Systems Division.
Allegretto took the company public to pay off Condec and Joe had no choice but to go along. The proceeds paid all indebtedness and the remaining $6M provided working capital for Unimation – not much for a pioneer whose technology was becoming obsolete, and with significant competition.
Enter Westinghouse: the beginning of the end
With the success of taking the company public, Allegretto told Joe that if he weren’t made CEO he would quit. Joe accepted Allegretto’s offer to quit and then reassumed the reins.
But things did not go well for the company. Sales dropped as the competition gained momentum. GM was leading the charge for electric-driven robots, which Unimation still did not have. Joe’s previous confrontation over hydraulics with GM’s CEO didn’t help, and between Cincinnati Milacron, Asea, and GMF Robotics, Unimation’s position was seriously undermined.
Yet Unimation had a number of suitors, all with a desire to gain a foothold in robotics. In December 1982 Westinghouse paid $107M, buying its way to the top of the domestic robot industry. The merger of the two companies moved rapidly.
Almost immediately, Unimation’s fortunes plummeted and its market share eroded. The general economic recession [1980-1982] caused a drop in sales, but industry competition contributed greatly to Unimation’s financial downward trend. The largest segment of our business, auto manufacturing, was shifting to electric-driven robots.
Goodbye Shelter Rock Lane
Those of us who grew up with the entrepreneurial spirit of Unimation could not readily adjust to the ponderous ways of giant Westinghouse. I was the first to go among the executives.
In July 1983 I became SVP of Robot Systems Inc. in Georgia. Joe continued as president of Unimation, though his was an association of oil and water. Unimation West eventually broke away; Kawasaki and Unimation Europe ground to a halt under the new management.
Finally, in 1984 Joe threw in the towel, saying, “I resigned in dismay. I was heartbroken because this was my baby, and it was crumbling before my eyes.” In 1985 he founded Transition Research Corp., which later became HelpMate Robotics.
In March 1985, I returned to Unimation as Manager of Distributor Sales. As I became familiar with the new Danbury operation, I felt that it was only a matter of time before Unimation would be absorbed into Westinghouse’s Pittsburgh operation. Sure enough, my duties were suspended as operations in Danbury wound down.
Westinghouse’s disclosure that it was setting up a joint venture with Matsushita Electric seemed the final move in a series of strategic shifts since the merger that were “essential to improving competitiveness in the factory automation market.” Time would show otherwise.
Before shutting the doors at Shelter Rock Lane – where it all started back in 1954 – and moving what little remained of Unimation to Pittsburgh, those eligible for early retirement were so advised by Westinghouse. I was one of them.
In 1989 Westinghouse sold Unimation to Staubli of France.