Joe Fuller: Most modern factories are neither dirty or dangerous, nor are they automated and lights-out operations. Intelligent machines allow companies to meet the demands of mass customization and high product variability. What are the implications of the emergence of cobots and associated technologies for the future of work? How will they help companies manage the demographic transition to an older workforce? And how can companies take advantage of these new technologies without writing off their massive investment in previous generations of industrial automation?

Welcome to the Managing the Future of Work podcast from Harvard Business School. I’m your host, Harvard Business School professor and visiting fellow at the American Enterprise Institute, Joe Fuller. My guest today is Patrick Sobalvarro, president and CEO of Veo Robotics. Patrick joins me to discuss the evolution of manufacturing and supply chains, as well as Veo’s advanced vision and sensor technologies, which promise to revolutionize robotic control with important implications for workplace safety and the future of work more generally. Patrick, welcome to the Managing the Future of Work podcast.

Patrick Sobalvarro: I’m delighted to be here. Thanks, Joe.

Fuller: Patrick, a lot of people who study the future of work subject are already paranoid enough of what the implications of robotization and other forms of automation are going to be in terms of the capacity of people to make a living. But could you tell us a little bit about Veo and, specifically, the idea of a collaborative robot, a “cobot?”

Sobalvarro: We make a vision system, a 3-D vision system, that is intelligent. And what it adds for robots is the ability to be responsive to humans in their presence. That responsiveness is really for the purpose of safety. And so collaborative robots are a technology that emerged about 10 years ago. At the time, I was the president of a company called Rethink Robotics. We were one of the first collaborative robot companies. And the advantage of a collaborative robot is that it is safe for humans to get close to. There are about 3 million industrial robots in factories and logistics around the world. And those robots are mostly very big robots that move very fast. They’re pieces of machinery that handle heavy things and accomplish tasks like welding and palletizing and so on. But they are mostly locked up behind cages, and interaction with humans is impossible for safety reasons. What we did at the beginning of the collaborative robot revolution, you might say, is introduce a new class of robot that was safe for people to be close to because they were so light and so a weak in terms of how much power, or force, they could apply to something, that if they bumped into a human, they could just stop, power off, and the human wouldn’t be injured. That’s what power- and forced-limited collaborative robots, which are all the collaborative robots that are out there now, do. That’s their principle of operation.

Fuller: When you say that a robot had to be in a cage, and that was a reflection of its interaction with humans, can you tell me a little bit more about what’s going on there for our listeners that don’t have a vision of what an auto plant or an aerospace plant looks like?

Sobalvarro: Not all manufacturing is done in an assembly-line fashion these days, but it is generally divided up into process steps. And those process steps for, say, a car might be, put the doors on the car or put the seats in the car or the instrument panel or something like that. With a dishwasher or a refrigerator, they might be somewhat different steps. But those process steps are discrete. And there’s also continuous-flow manufacturing, where you’re making something like gasoline, but we don’t really address that. But in durable goods, and in a lot of consumer goods manufacturing, you’ve got these individual process steps. And some of them are performed by robots. For example, almost all welding in manufactured goods is performed by robots, and things like dispensing glue, just things that require a lot of constancy and a lot of reaching and so on. Those things are better done by robots than by humans. But then there are a lot of process steps that are impossible to do with robots, and those are ones that require judgment or a lot of dexterity on the part of the production worker. And so judgment might be a quality control step or deciding how to adjust something in final assembly. Quality control might be just checking that things have been put together right or doing a measurement or whatever. What you see, for example, in final assembly for a car is there’s a first set of steps called body-in-white; it’s welding and painting. And those are all done by robots, and people aren’t in there until the very end of the paint line, where they’re touching things up and so on. After that, you’ve got maybe another 300 process steps that are almost entirely manual. And people are moving around things like those doors or those seats with lift-assist devices. The cycle for each process step is about 40 to 50 seconds. It gets very exhausting on a shift, and workers have to be cycled through process steps so they don’t get repetitive-stress injuries. The caging of the robots is necessary because a robot, unlike what you might see on a TV show or in the movies, robots aren’t smart. They’re just machines that move from point A to point B. So you’ve got to be careful.

Fuller: What’s the value of creating a system like you’re doing at Veo that allows these big FANUC or ABB robots to be cage free, like a chicken, I guess?

Sobalvarro: Well, if you think about the production labor workforce in really all major manufacturing economies, what you see across the board is that workforce is getting older. In some European countries, you’re getting average ages in the high 40s, low 50s. And so the elements of the process step that involve moving heavy things, you might do that with a lift-assist device—and a lift-assist device is sort of like a manually operated overhead crane, but a small one. What we can do, what our system does, is it makes those big robots capable of being close to humans and being responsive to their presence, so it will slow or stop them if they are moving too close to a human. And then the robot can position that instrument panel in the vehicle, for example, and the human worker can jump in and connect the wiring harness or drive the fasteners or do whatever requires human dexterity and judgment. And the robot is doing the heavy lifting. Another example might be positioning something at the ideal angle for a human worker to do work that is ergonomically optimized.

Fuller: Certainly one of the things we’ve seen here, at our project at Harvard Business School, is that it’s been harder and harder for manufacturing companies to attract the interest of young workers. And a lot of people entering the workforce or their parents or caregivers or their advisers think of manufacturing still as a dirty, dark, dangerous undertaking that may not be one with much of a future. It sounds, though, like this is, in addition to addressing the physical aspects of work and reducing the prospect of injury or poor performance due to a lack of precision, that it also probably leads for a more-efficient design of work and has good economic rewards for the employers’ deployment.

Sobalvarro: We often say, “We’d rather not have a company than do something unsafe.” But the primary economic imperative for a customer of ours is increasing productivity. And so, if you can reduce the time that it takes to do a process step, that marginal increase in your productivity really helps the bottom line tremendously. And as we know, productivity increases are what lead to things like having a lot of teachers or a lot of health care workers. If you think about the United States in the early 1900s, there was a huge portion of the labor force that worked on farms. And there’s a much smaller portion of the labor force that works on farms today, but we sure have a lot more health care workers and teachers, and that’s a direct result of increased productivity in agriculture. There’s also been increased productivity over the decades in manufacturing.

Fuller: What inhibits the deployment of technology like this? Because it sounds like a pretty compelling business case to deploy it.

Sobalvarro: One thing that is really key—and that a lot of people might not be aware of—you think about dirty, dark and dangerous, as you were saying earlier, factories. Those factories, I’ve been to hundreds—literally hundreds—of factories in the United States over the past few years and before that, because I’ve been in robotics and computer vision for quite a while. Those factories are not actually very dirty or dark. And in fact, they are much less dangerous than, say, crossing a street. If you look at the safety record of robotics, you’ve got several million, about 11 to 13 million people going to work in a factory in the United States—at least in pre-pandemic times—every day, and hundreds of thousands of robots and other kinds of automatic machinery. Well, over the past about 35, 36 years, since 1984, when OSHA began collecting these statistics, there’ve been 40-odd accidents involving robots in US factories and 14 or 15 fatalities, which is too many. But compare that to 30,000 people killed in automobile accidents per year. There’s a very fanatical safety culture in industrial automation. One of the impediments to the adoption of a technology like ours is we have to seek functional safety certification from an internationally recognized body. We’ve been working with TÜV Rheinland for several years. And it’s a little bit like making a medical device. It’s something that requires a tremendous amount of documentation and processes, failure mode effects analyses. All of our hardware is dual channel. If you follow the reauthorization of the Boeing 737 MAX, many of the things that allowed them to return to the skies had to do with dual redundancy. Our system is built like that throughout. Getting that functional safety certification has been an impediment to quick adoption, but we absolutely wanted it and needed to do it. And we’ll have that in February of 2021, so we’re really looking forward to that.

Fuller: That’s exciting. And of course, OSHA is the Occupational Safety and Health Administration, part of the US Department of Labor that monitors workplace safety. Patrick, what about retraining the worker to accommodate this technology? One of the things that we’re hearing very regularly is that, as new technology gets deployed, those incumbent workers, as you described them, in a lot of these manufacturing settings may be a little bit older, relying on lots of years on the job to be productive. How do you get that incumbent workforce comfortable with working with this new technology to get the most out of it and to extend their careers? Because a lot of what this technology does is to perhaps prevent them from getting repetitive stress injury or otherwise getting so fatigued on the job that they decide to retire as opposed to keep working at it.

Sobalvarro: Yeah. I would split my answer to that question into three different kinds of workers. One kind of worker that we rarely think about in the United States is the industrial engineer. There are many, many industrial engineers in the United States. They are kind of the spiritual descendants of a skilled worker called a “millwright” in the 19th century—the people who had to figure out how to put together factories. Their descendants today are industrial engineers. And these are typically people with four-year degrees in mechanical engineering or electrical engineering or sometimes that program is called “industrial engineering.” And they’re responsible for building the machinery and planning the process steps that go into making all the things around us. Those workers, there aren’t enough of them. There are a lot of them in the United States, but they generally are educated in the middle of the country, and they generally find their jobs in the middle of the country. There really aren’t enough people being attracted to this really creative, ingenious work. I would agree with people in the trade organizations, in manufacturing automation, that we need to wake people up to the existence of those jobs, which are highly paid and really creative, wonderful jobs. There’s another tier. As you increase automation, you need more technicians. And these technicians may be, for example, machinists making an end effector for a robot. And these are highly customized, generally, to the thing being done. They’re typically not mass produced. Or they may be making a fixture for a part or something like that. You also need electricians who are responsible for wiring up this power safely—people who understand things like compressed air and pneumatics and so on. There’s a lot of technician work. This is typically fulfilled by people with either extensive on-the-job training or with two-year associate’s degrees from community colleges. And then, finally, there are production workers. We have a shortage in all three categories.

Fuller: Patrick, if you listen to talking heads on the all-news stations or you read newspapers, a very common trope is that robots are coming to displace workers, and robots are going to do everything and people are going to have to find something else to do. But what you’re describing, of course, is not a shift to a process that’s lights out, no human involved, but just a different type of human with a different type of skill set being in more demand as a function of the type of automation you’re talking about.

Sobalvarro: One piece of research I’ve followed is this question of what happens to jobs as automation increases. And I think there’s a popular notion of the lights-out factory. My cofounder, Clara Vu, our chief technology officer, likes to say, “The lights-out factory is a dead end.” And the reason for that is the rate of change in product cycles has increased so tremendously. You used to make the same thing again and again and again for decades. That’s no longer true. And you can’t amortize the cost of full automation over the number of units you’re going to build. Elon Musk famously tried to build a fully automated lights-out factory for making Tesla Model 3s, and after spending billions of dollars on that, he had to stop and built a factory in a parking lot that looked an awful lot like other car factories. And he said, “Humans are underrated” in a tweet about that.

Fuller: He did.

Sobalvarro: Well, they were underrated by him, but I think the rest of the automotive industry had understood that decades ago.

Fuller: It’s nice to have the capacity to drop a few billion here and there and not feel it at all, which Tesla’s been able to do with its valuation. Patrick, how do you think this affects supply chains over time? Any thoughts about the theme of reshoring? Are the types of technologies that you’re pioneering at Veo or that you see on the horizon going to change the way companies think about their supply chains and lead to that often mentioned but seldom observed industrial Renaissance in the United States?

Sobalvarro: Yeah. Manufacturing GDP hasn’t really declined much in the United States. One thing that has happened is that it’s grown a great deal in Asia over the past few decades. The term “reshoring” I think is often used in kind of a political context. But if you look at it from an economic point of view, there are always reasons to move manufacturing closer to the point of demand. And those have to do with just being able to get products designed and delivered to customers more quickly and less expensively. If you’re just looking for low production labor costs, you also suffer from a longer supply chain. Things have to move by container. It takes weeks for a container to be shipped across the ocean. You’ve got these transnational boundaries in different time zones of design and test and stuff like that take longer. And localization is harder for products. The idea of a single global product doesn’t always work out, especially in durable goods. If you think about what’s happened over the course of the pandemic and the trade conflict, you’ve had a series of supply shocks that have hit manufacturers, and they’ve simply accelerated this trend to move manufacturing closer to the point of demand.

Fuller: Well, so you’ve been in industrial automation for your entire career. You’re leading one of the very innovative companies that’s going to get the technologies that are already deployed and make them more productive. If you can dust off your crystal ball, how do you see manufacturing evolving in the next five, 10 years? And what types of technologies are on the horizon that you’re excited about?

Sobalvarro: Well, there are macro trends that I think have driven a lot of the change in manufacturing. The macro trends are the greater availability of products from around the world that, to a great extent, is driven by the communication that the Internet allowed. Somebody can look at a product—say, a kitchen appliance or something like that that’s available in Japan—and say, “I could really use that here in New York.” That has opened up opportunities for a lot of new companies to start up production of things. It’s driven demand experienced by legacy companies to vary their products. You’ve got much shorter product cycles as a result because of all this competitiveness. As consumers, we are enjoying lower prices. We’re enjoying greater choice. Quality demands have also gone up. I’m old enough to remember the days when somebody would buy a new car, and they’d have to wait six months or so to buy Consumer Reports to find out that their new car was a lemon.

Fuller: Yeah, you’re getting pretty close to the nerve there for me. I think I am that person.

Sobalvarro: Yeah. Well, when you think about it, there aren’t that many lemons anymore. The quality of stuff that you buy has increased dramatically. And I think that’s driven by consumer communication. But those trends place increased pressure on manufacturers to be very, very flexible—to be able to change their processes very quickly, to be able to ramp up the building of a new product very quickly. And that’s why we say that production labor jobs are not going away, because you can talk all you want about artificial intelligence—and just for the audience’s background, I got my PhD in computer science at MIT, and I did my master’s in the artificial intelligence lab and worked as a research staff member there—the claims of solving general artificial intelligence are terribly overblown. And there is no replacement on the horizon for the ingenuity and flexibility of a human production worker. We don’t see those jobs going away. But what we do see is a need to have automation help manufacturers be more flexible and more responsive to quality demands, to shorter product cycles, and so on. I think we will see more automation. Certainly, we’ll see increased use of machinery and also software as you go further back in the design process. But I don’t think that what we’re going to do is progress toward the lights-out factory that Elon Musk wanted to build.

Fuller: Patrick, one of the origins of our project, Managing the Future of Work, was a belief by actually alumni of the Harvard Business School that we were surveying very regularly that the US had historically had an advantage in terms of a skilled workforce, and that advantage had waned. And, in fact, that the availability of skilled workers was now a decrement to American competitiveness when compared to global competitors. How do you view the health of manufacturing workforces across geographies?

Sobalvarro: One thing that I have observed—I’ll give you a little bit of history about Veo Robotics. Our seed investment came from Siemens, the biggest industrial automation company in Europe. And while we do also have investment from blue-ribbon Silicon Valley investors like Google Ventures and Lux Capital, our most-recent investment round was led by Alpha Intelligence Capital—AI Capital—out of Luxembourg. We do spend a lot of time in Europe. And what we see in countries like Germany, for example, is a greater integration of manufacturing into the sort of popular conception of good work. I think the cliché here in the United States is that manufacturing jobs and manufacturing work is just a dreary stuff. And in Europe, that really isn’t the case. And, certainly, it doesn’t seem to be true in reality to me when I visit factories in the United States, that it’s dreary. It seems very alive, and those jobs do pay better than being a barista at Starbucks. I think there is, from top to bottom in Europe, you see vocational education all the way up to PhD-level studies that concentrate on manufacturing and manufacturing technologies. And I think that’s led to a kind of healthier pipeline of skilled workers coming into the sector in Europe.

Fuller: Certainly the longstanding and very well-developed program of apprenticeship that you see in a Switzerland, a Germany, a Denmark provides a supply of workers. And supply and demand get in balance, and all of a sudden, you’ve got a virtuous cycle. How would you contrast that with what you see in the industrial centers in Asia—China, Korea, Japan?

Sobalvarro: We also do have investment from both Japan and China. And so we’ve spent a fair deal of time in both places. I think in Japan, certainly manufacturing has historically—certainly over the past few decades—been seen as a national strength. And so we certainly see something a little bit more similar to what goes on in Europe versus the popular treatment. And, of course, China sees manufacturing as key to its future. Again, there is an ecosystem that exists in the United States that has evolved over the decades. My wife is the youngest child of seven. Her father fought in World War II, came back, and, with a high school education, became a skilled sheet-metal worker. And his union provides training at community colleges, two-year programs. That that sort of thing, unfortunately, has died out to some extent in the United States. Community colleges still fill a crucial need in bringing skilled technicians and engineers to the workforce. But I really do think, in order to maintain a healthy manufacturing sector in the United States, we need to support these institutions. And I would not minimize the effect of the cultural treatment of manufacturing as something that happens in the flyover states and a dreary occupation. I wouldn’t minimize that because these really are actually great jobs. They’re super fun. These factories are fascinating places, and I really wish we’d get beyond that in the United States.

Fuller: We’ve certainly heard that from more than one CEO in industrial sectors in the US, complaining that, between the way manufacturing is portrayed by everyone—from educators to social leaders and in the media—that it’s no wonder that aspiring workers are reluctant to pursue those careers.

Sobalvarro: There’s a systems integrator we work with in Minnesota who builds robotic systems the size of a football field. You go and you talk to them about it, and they’re like, “Yeah, we’re making that thing.” And it’s fascinating, ingenuous work.

Fuller: Well, Patrick Sobalvarro, CEO of Veo Robotics, thanks for joining us on the Managing the Future of Work podcast.

Sobalvarro: Thanks very much, Joe. It’s been a great pleasure

Fuller:We hope you enjoy theManaging the Future of Work podcast. If you haven’t already, please subscribe and rate the show wherever you listen to podcasts. You can find out more about the Managing the Future of Work project at our website hbs.edu/managingthefutureofwork. While you’re there, sign up for our newsletter.