Masters of Manufacturing: Joseph F. Engelberger
This is the fifth annual installment in an article series we call Masters of Manufacturing. In these articles, we honor a distinguished figure in manufacturing technology, and by doing so, we hope to remind readers that a career of great achievement in manufacturing is still possible.
By Patrick Waurzyniak
Often called the "Father of Robotics," Joseph F. Engelberger earned his BS in physics and later an MS degree in electrical engineering at Columbia University before embarking on his career, starting as an automation engineer designing controls for nuclear and jet engine applications. In 1956, Engelberger began development of an industrial robot, and he later founded and served as the first president of Unimation Inc., builder of the world's first industrial robot, the Unimate, that was initially installed in a die-casting operation at a General Motors factory in New Jersey in 1961.
Author of numerous technical articles and two books on robotics, Robotics in Practice and Robotics in Service, Engelberger in 1984 founded a new company, Transitions Research Corp. which became HelpMate Robotics Inc., with the mission to give robots a spectrum of sensory perception enabling mobile, sensate robots to work with human mentors in service activities. The company's first successful service robot, HelpMate, was a robotic hospital courier. In 1999, HelpMate Robotics was sold to Cardinal Health, which merged it into its subsidiary Pyxis Corp.
Recipient of several honorary degrees and numerous international awards of recognition, Engelberger recently retired from active management, but he continues to pursue his dream of developing a two-armed, mobile, sensate, and articulate robot that could be a servant-companion for elderly and infirm individuals who otherwise would need nursing home care. Engelberger lives with his wife Marge in Newtown, CT. In an exclusive interview with Manufacturing Engineering, Engelberger discussed his life, his work, and his vision for the future of robotics.
Manufacturing Engineering: When did you start thinking about robots?
Joseph Engelberger: I went to Columbia University under the auspices of the Navy V12 program, and I followed some years behind Isaac Asimov, who wrote the book I, Robot; incidentally, the movie doesn't do it justice. That inspired me, and I had it in the back of my mind that that was a great thing to do. The Navy made a physicist of me, and it was nice to get out of the Navy and have a whole bunch of companies wonder what a physicist was. So it was very easy for me to get a job, and I went into manufacturing, developing and manufacturing controls for jet engines and for nuclear power plants. The Navy had me from 1942-46 before they let me out, and I was working on the atom bomb project down at Bikini Island before that, so I had a lot of background.
ME: Were you a fan of Isaac Asimov's writing?
Engelberger: I was a big fan and years later, when we had our robot company going, I invited him in to speak to our employees. One of the things he said was, 'I wrote these stories to pay for my college education. I didn't think anybody would really do it.' He was an inspiration, and then a peculiar thing happened. I was in the business of high-tech nuclear controls and at a cocktail party, I met a guy named George Devol. He had a patent, called Programmable Manipulator; I looked at it and said, 'You know what? That sounds like a robot to me.' So I had the company I was with then take a license under that patent. My position was, 'If I look at what we do in controls today, we have the technology to do this. We could make an industrial robot.'
ME: This was in 1956?
Engelberger: Yeah. At that point, in the company I had a few people working on what the market conditions could be, and I think that's very important, because other companies that would try to compete later on failed. We visited 15 companies in Bridgeport, Connecticut. We visited General Motors, Ford, and Chrysler, and came up with what this thing should do to be satisfactory in this industry—and out of that came the design of the Unimate.
ME: What company were you working for at the time?
Engelberger: Manning, Maxwell, and Moore. Mr. Moore, who was my immediate boss, came to my defense, and I was able to get money to buy a little operation in Danbury. I got outside directors and convinced them to give me a chance to raise some money. They gave me three months to raise the money, then after that I took over the license. We didn't have the capability to do the robot thing anymore, so I got the license from George Devol. We started modestly to get the development going again. It wasn't until 1961 that we had a machine ready to go and working. We placed it in a General Motors plant in Ternstedt, New Jersey, on a die-casting operation.
ME: How were robots integrated into factories?
Engelberger: If you recall, we picked male labor and some jobs that took two hands to do. We weren't making two-handed robots—we were making one big strong arm. So it had to lift heavy things, and it had to go more than one shift to pay off. That means the ideal thing was a job involving major capital equipment. The first job was a die-casting machine. Here's a big piece of equipment; one wants to run it around the clock, because you paid a lot of money for it. The parts were only about 30 to 40 lb [13.5-18 kg], but hot as hell. They came out hot from the die-casting machine, then they have to be quenched and put into a trim press. A tough, miserable job for a guy. We used to say, 'The jobs that we want are the hot, hazardous, and boring.'
ME: Dull and dangerous jobs. What were the characteristics of that first Unimate robot?
Engelberger: Well, it was hydraulic, first of all. It didn't have an arm like a human being, it had an extending kind of an arm. The advantage with an extending arm is that it can go into a die, and it can go into a machine tool. It was very strong, because it was hydraulic, and it also had a pretty good brain. You know, the machine that went into General Motors, after 100,000 hr, it ended up in the Smithsonian, the first industrial robot.
ME: When you say it had a pretty good brain, are you talking about the processor itself?
Engelberger: One of the big things we did was we went solid-state electronics on our control. At the time, a transistor cost $1.70 each, and people who tried to compete with us, like AMF, went with radio tubes, which cost 22 cents each, but they didn't have the capability we could get with a transistor. It was a solid-state controller, composed of transistors. We were saying 'It's got to get cheaper,' never knowing it would be a fraction of a penny per function, eventually.
ME: So the first machine was revolutionary in many ways, even though it might have had some limitations?
Engelberger: Revolutionary because no one else had it. Some people like AMF and Cincinnati Milacron, who tried and who failed, by the way, and others took licenses from us. General Motors did, IBM did. We had that first patent I told you about, and a few others in between, but these newcomers never really approached the market in a logical way. Even today, I keep telling people, 'You've got to remember the cost-benefit ratio.' Now if the cost is not less than the benefit, it's not going to sell.
ME: Was programming a problem for adoption of the first robots?
Engelberger: No. Programming is always a problem, but if you have bright engineers working at it, they'll get through that kind of a problem all right. I didn't think it was that much of a problem for factory work. At the time, the public was interested in the robots that were going to be part of our world. Today, you hear 'What about personal robots?' Back then it was all there, and we went on the Johnny Carson show. It was great fun. It had to be done in California, because the studio here wasn't big enough for the robot, so we went to California. And the theme of it was 'A Robot Could Take This Show Over.'
The first thing we had to do was putt a golf ball, to see if we could putt a golf ball better than Johnny Carson. It took a while for the robot to warm up, and it was noisy too, so I had the robot address the ball, like people do. That was warm-up time, and then finally, it tapped the ball into the cup, people clapped like mad. I handed the club to Johnny, and he said 'No, the robot'll beat me.' The next thing we did was the beer commercial. We had a beer can there, and I cheated a little bit, I had the tab on the beer can open slightly. The robot pulls the tab open, grabs the can, and pours the beer, and everyone was happy. The last thing I did was, 'We can lead the orchestra as well,' and showed people how I programmed it. I gave it a baton, led it through some motions, and the band played along with the motions of the the robot. At the end, the robot throws the baton down, grabs the band leader's accordion and smashes it down. Of course, it was a substitute accordion.
ME: So at the time, the public perception of robots was how robots could affect everyday life?
Engelberger: That's what people wanted to do. I was on show after show, where they wanted to know what it could do that would help human beings.
ME: I understand the first reference to 'robot' was in a 1920 Czech play?
Engelberger: Rossum's Universal Robots, or RUR, by Karel Capek.
ME: And the term 'robotics' was coined by Asimov?
Engelberger: Yeah. And you remember, the distinction here has not gone well. Rossum's Universal Robots was a play that's sometimes still acted on college campuses, but the premise was that two countries build robots to fight their wars so they don't have to go out and kill each other. But the robots got smarter and said,'What the hell are we doing wiping each other out, while these people are sitting around,' and they took over the world. This has been a theme of science fiction now for a long time—the robots take over the world—and even today, you can't get people on television, the press, without bringing up the whole issue, 'If we make them smarter, they're going to take over the world.'
Asimov's big pitch was, 'I gave the robots the three laws of robotics: They cannot harm human beings, or through inaction, allow human beings to come to harm.' That's Rule 1. Second rule, they must always obey human beings, unless that's in conflict with the first law. So a robot can't say 'Kill that guy!' And the third law is the robot must always protect itself from harm, unless that's in conflict with the first and the second law. In his stories, he put the robots in dilemma situations—that's part of the theme of the stories. Now I take that as very important. In 1956, the same year that I started, a movie came out called Forbidden Planet. It's a science-fiction movie in which there's a robot called Robby, Robby the Robot. In one scene, Walter Pidgeon gives Robby a gun, a blaster, and orders it to kill someone, but the robot screws up and his wheels go wild, he can't kill. Today what disturbs me is that the military is very busy now trying to build robots for combat. Not just robots in the air and underwater, but robots that are anthropomorphic, that would go into a building with enemy in it and attack them. The biggest source of R&D right now for robotics is with the military. And if you believe that we'll be the only country with them, you're wrong. I still remember a quotation from Einstein when he said, 'I wish I never signed the letter.' Now everyone is going to have nuclear power. They're not going to stop Iran or North Korea. And everyone will have robotics. They'll have robot warriors, and RUR may come to pass.
ME: What were some reasons for robots catching on in manufacturing during the early years? Was it improved computers or programming, or even tactile capabilities?
Engelberger: There's so many things we could do without that tactility. We had the one machine in General Motors, and then we had other machines in die casting and machine loading. And then General Motors really came along, gave us a big order for 36 robots, for Lordstown. It was the first major order, and that publicity stimulated so much. It stimulated the Japanese. Here's General Motors, buying 36 robots all at once, and that got me to be a guest of the Japanese government.
ME: You visited Japan in 1966?
Engelberger: Yeah, and it was something. At the time in the United States, I had trouble getting 6-10 people to listen to me about what I wanted to do, but when I got to Japan, they had a conference with 400 executives wanting to hear about robotics. I've been to Japan 38 times since.
ME: Why were the Japanese fascinated with robotics?
Engelberger: A lot of things. First of all, there was RoboBoy in their television, it had been on for a long time. They had no unemployment. They did not like guest workers. They didn't want to bring foreign help in, so the robot comes along and they saw it as a way to improve their manufacturing skills. The Japanese formed the Japan Robot Association three years before we had one in the United States, and they got the president of Mitsubishi to be the first president. We begged General Motors, we begged everyone to give us the first guy to be the president, but we couldn't get them interested here in this country.
ME: Why were US manufacturers slow to adopt robots in factories? Weren't the early Japanese efforts mostly nonprogrammable machines performing automated tasks?
Engelberger: That's automation. At the outset in the United States, I suggested that we make robots a subset of automation. That to me was OK. Don't emphasize this walking, talking doll and all those things. It's a subset of automation, programmable automation, but our guys wouldn't have any part of it. No, the Robotic Industries Association when it was getting started said 'We want to be something separate.' It was a subsidiary, at the time, by the way, of the Society of Manufacturing Engineers, which supported the Robotic Industries Association when they didn't have any funds of their own. This was in the early 1960s.
ME: Were you surprised at this kind of reluctance by the US manufacturing giants?
Engelberger: Yeah, look at the fact that they were so scared of labor, that there's going to be people writing articles all over 'The robot's going to take all the jobs away.' Not so, the robot had a big advantage, it enters the workforce gently. You don't displace mobs of people all at once with robots. Now if you automate, sometimes everyone goes—it's full automation then. And when I talked to the labor unions, the guys who run labor unions were not dummies. They said, 'Look, we're not against robots. We think the jobs they do are miserable jobs anyway, and our workers shouldn't do them. We don't want you to displace our labor en masse. If they displace them slowly, it's acceptable.' The way it works, the robots go in gently, they do not result in masses of people going out at once.
ME: What were some early difficulties making robots?
Engelberger: When we started, from a reliability point of view, we had a 400-hr mean-time-before-failure. I used to tell customers, 'You know what, that's about the same as a human employee—he goes down once in 300-400 hours.' We got away with it, and we had good service. I said, 'Here's the thing you've got to realize is that, if something happens at the end of the second shift, we'll get to you and fix it before the beginning of the first shift. If something happens at the beginning of the first shift, you're going to lose that shift.' That's where I came up with 400-hr MTBF, and the amount of downtime.
ME: So were you able to fix it after the second shift?
Engelberger: Sure. We had good service at the time and we were able to train the customers to do it. Today, most industrial robots have a 10,000-hr MTBF—a tremendous improvement. Four hundred hours was enough at the time, but that's where the technology had to improve, we had to find ways to do it better.
ME: How did it improve to 10,000 MTBF?
Engelberger: Later when selling industrial robots, I used to tell people 'Look, I'm going to sell you a mechanical robot design that's going to be very reliable. I'm going to give you the electronics, and I'm going to rent you the software, because electronics has become so cheap.' It wasn't an issue any longer. People don't understand that. The chips are negligible in cost. You put it together, get value out of it, but you better do a good electromechanical job on the physical robot. That's where the real design is—you need damn good mechanical/electrical engineering.
ME: How much did the first robots cost?
Engelberger: At that time, we were selling the robots for $25,000 apiece. After we started Unimation, they were costing us $60,000 apiece to manufacture. Our investors realized that didn't sound like good business, so they hired McKinsey, who did an analysis, and said, 'If you make 200 of these a month and you do value engineering and you tool them, you can make them for $8000 apiece.' Investors had to have faith that we could do that. We get to 200 a month, and what we were doing was meeting the market. We had to believe. We had a faith that the $60,000 isn't going to last, but when we have quantity production, the price is going to go down.
ME: So value engineering helped you find the sweet spot, the price at which you make your profit?
Engelberger: Well, if it turned out that you couldn't make this machine for something well below what someone's going to pay for it, you don't have a product. We still kept the $25,000 price. We actually got around to doing something else, which is important. We started to rent machines. We would say to someone, 'You're going to pay $6 an hour for the first shift, and $3 an hour for the second shift. You don't like it—fire it, like an employee.' If a guy in management is worried about his career, he thinks 'If I buy these damn things and then I stumble, it's no good, I'm going to lose my job.' But this way if it doesn't work, you fire it. And no one ever fired them.
ME: Nobody ever fired them?
Engelberger: Never sent them back. They were so much cheaper than the labor that they had. You rent it by the hour. They went along with it and, of course, we ended up having cash cows out there that were delivering money to us all the time. We kept it up, by the way, with HelpMate. It was much more important with HelpMate, if a hospital administrator is on the flypaper for buying a robot in the hospital, he's not going to do it. But on the other hand, if he hired it, and didn't like it, he'd fire it.
ME: What is your vision for a mobile, sensate robot to work in the healthcare industry?
Engelberger: What I'm after now is to develop a twoarm, mobile, sensate, articulate robot that could cook, clean, help with ambulation, handle security, and be a companion. It doesn't exist today. The technology is available, just like it was for Unimate back in '56, but yet I can't convince anybody to put their money into this. What I need right now is to prove to a lay person who's not an engineer that it can be done, and I need $700,000, of which I'll put up $200,000 myself, and I can prove that we can go ahead. Now, that won't give you the product yet—you still have to do what I did with Unimation. You have to do value engineering, you've got to do tooling, and that will be a $3 million total investment, and it's going to take a total of 27 months, with the first phase taking only 14 months.
ME: Describe a mobile, sensate robot.
Engelberger: Well, there's three things: one it's got to be mobile. Now remember, HelpMate was mobile. Second, it's got to be two-armed, because you can't accomplish the things that need to be done in the household with only one arm. It should be sensate; we're sensate. In HelpMate, we used both vision and ultrasound. You need multiple sensors, ultrasound and tactile. Ultimately, it's tactile sensors; in other words, you don't bang into anything, because you have a tactile sensor that stops you. In the hand gripper, you want tactile sensing. There are two kinds of gripping in the hand. One is forcesensing. If I put my hand on the table like this, my wrist tells me I have pressure there. But if I grab the glass, I want tactile sensing. Now when I go to the refrigerator, I open the door, and even that's a tricky thing. When you open the door, you don't understand that what you do is let some articulations go limp. You don't drive the door open, you grab the door and pull, and then the limp ones let go and the door opens. Now you look into the refrigerator and you have to sense what it is I'm trying to find— what's the difference between milk and beer? You know what that is, visually. Now the next thing that happens, when you reach in there and reach for it, your hand occludes the vision signal. Now you don't see any more. What's next? You grab the beer can or the milk bottle, and you sense it, that's tactile sensing, so you use vision and tactile sensing to get to it and bring it back.
ME: Are there technical hurdles combining sensors?
Engelberger: No. Good engineering, that's all.
ME: So most key technologies are there, as sensors have improved a lot, with better tactile-sensing devices. What do you see in the future?
Engelberger: Additional sensory perception. Better sensory perception than exists. It's always being improved. I said there's enough to do the job, but boy, we can get better with sensory perception. That doesn't mean sensors can't be improved. Better end effectors. The hand, the end effector on a robot, is not very good right now.
ME: Is that also true with industrial robots?
Engelberger: Always helpful in both cases.
ME: And computers and processors are so advanced it's not an issue?
Engelberger: It's so advanced. It's enough, it's not an issue.
ME: What about programming?
Engelberger: Well, there are programming techniques that make it easier, but I would expect that a handicapped person would program the robot by voice.
ME: It's difficult for engineers to create a new industry; you really have to think outside the box, don't you?
Engelberger: It's hard, and I'm already too old to do what I did with Unimation, visit 46 companies to find someone who's going to be a good fit for me. I haven't found anybody to pick up the gauntlet yet. I have some good people who are willing to try.
ME: You haven't fully retired?
Engelberger: I'm still lecturing. I've been to Korea a couple years ago, I went to Japan last year, and I'll probably be going to Germany next month and I'll still make the pitch.
ME: Do you recommend robotics and manufacturing to young people today seeking a rewarding career?
Engelberger: I would recommend robotics to anyone to start. Remember, I started young, and this industry is still young—and there's so many opportunities. You know, from my elbow to the end of my hand, there's 30 articulations that I can mentally control, and a few that I can't control. There's a challenge still to come up with so much more. What evolution, or if you want, God has produced, is not easily replicated, so there's going to be opportunity all the time. So it's not a dead-end.
ME: Has the robotics industry lived up to what you thought it would be?
Engelberger: We have an $8-billion-dollar industry worldwide in robots—that's much too small for what it could be. If we had the service robots aligned to what they can be, it'd be up there around $100 billion, where it should be.
ME: Do you think a sensate robot will be in the market in 10 years?
Engelberger: Well, I said we'd be in the market in three years, if we'd start today. I have to tell you that I also said that in 1990, when I wrote Robotics in Service that it could be done. The question is, it can't be done unless you do it.
ME: Are there technical hurdles?
Engelberger: No. The market pressure is growing in the whole world.
ME: Who could do it?
Engelberger: The only ones that could do it are the Japanese, the Koreans, maybe Electrolux in Sweden. Kuka is a possibility, because they have tried some unique things. The ideal company would be a healthcare company, someone within the business already handling assisted-living who can see the advantages of robotics. The Department of Health and Human Services expressed a lot of interest but said, 'We don't do R&D.' Now, we did some R&D on it with NASA, but NASA's interest wasn't really in that, they just thought they would get something useful in space. I don't blame them, that's OK—I think space is a great place for robotics.
ME: How much robotics is being used in space?
Engelberger: All the Mars Rovers were robotic. And they're not very sophisticated, by the way. Much more sophistication could be put in those robots, if they wanted to. You know, the first one up there came up against a rock and couldn't go any further. On the other hand, the last three of them are still alive! They're still poking around a bunch of things. The President has a number of new objectives. One is a Moon landing again, and another is eventually humans to Mars. Most of the projects have been downsized, because there's no money.
This article was first published in the July 2006 edition of Manufacturing Engineering magazine.