When the U.S. globalization push of the 1990s outsourced manufacturing, proponents could not have known the full impact this shift would have at home. Not only were many towns wiped out, their commerce to be replaced by service industries, but the manufacturing sector itself would struggle with the loss of infrastructure — buildings and machinery — and with manufacturing intelligence as well. Manufacturing intelligence, the kind of know-how and workers that we need to take back industries, was lost.
The country’s shrinking manufacturing base benefited many other global economies, however. China, of course, became a major trade partner. But China, which is currently our second-largest trading partner, our third-largest export market, and our biggest source of imports, per a September 2015 Congressional Research Service report, was not the only recipient of the global economic boom.
Further, as China continues to build for growth, the U.S. continues to redefine global logistics and rebuild infrastructure with expansion of a few East Coast ports and the enlarged Panama Canal (through which China sailed the new canal’s first container ship). Seven of the world’s top 10 container ports are in China.
All of which adds up to big continued challenges for the U.S. as a global manufacturing and logistics leader — infrastructure, software integration, and especially the labor force have some catching up to do. The U.S. now faces some big manufacturing challenges, and we’re going to need serious brainpower once again — the kind of human and machine intelligence that took us to the moon and beyond — to address them.
The Big Challenge: Automation
When companies bring products back from off-shored manufacturing operations such as China, the processes cannot be simply dropped back into place and configured with the same old machines and process flows from 20 years back. Most re-shoring operations include new and upgraded equipment, software and, in some cases, advanced automation that includes collaborative intelligent industrial robots.
Robots, says Jim Lawton, chief product and marketing officer at Rethink Robotics, offer the tremendous competitive advantage. I talked with Lawton about the pace at which American companies are adding new robotics and automation, as well as the kinds of automation he sees coming on.
How fast can we bring in the new robots? Some early adopters feel that we can’t move fast enough, while others are finding that mixing old machines with new tech is too disruptive for on-going production. And then there is The Fear Factor …
Lawton is excited by the whole robotics movement, although it presents intimidating choices and decisions. “I think that as we’ve seen with many other technologies, people have become skeptical about whether the new systems will work as advertised. That’s healthy because lots of robotic solutions have been not-ready-for-primetime. People have all sorts of questions, and of course there is the fear factor — ‘Will I get fired if deployment doesn’t work?’”
The decision to deploy robotics requires in-depth knowledge of the current production process as well as a comparison of machine capabilities with upcoming new product development plans. Companies like Donnelly Custom Manufacturing’s Committee believe that, and some adopters choose to slice off a small piece of production where intelligent machines can be cleanly inserted and managed by employees ready and eager to work with the robots.
Lawton has advice for executives contemplating their robotics technology future. “It’s important to understand where the technology will be applied, and where it does work and where it does not work. All these points lead to companies asking questions because they are thinking, ‘Hey, there’s something real here, so what would it take to be successful?’ and more importantly, ‘do I have a choice?’”
Lots of robotic solutions have been not-ready-for-primetime. People have all sorts of questions, and of course there is the fear factor: ‘Will I get fired if deployment doesn’t work?’"
— Jim Lawton, Chief Product and Marketing Officer, Rethink Robotics
But the new age of industrial collaborative robots is just beginning. “The book hasn’t been written yet that will provide all the critical answers,” Lawton says. “We are in the phase where now a lot of company projects are deploying collaborative robots for more progressive industries. These pioneers are building the knowledge base about when to use robots, how to think about the technology, and what should be a robust manufacturing strategy. Even the idea of developing a manufacturing strategy can prompt many questions.”
But we do have some very good examples of hardcore industry applications of the new generation of collaborative robots. “The two that come to mind most quickly,” Lawton says, “are the automotive sector with lots of experience — 65% of all robots in the U.S. are sold into automotive. Automotive companies are not afraid of robots or making the investment.” The second sector moving into collaborative robots for manufacturing is high tech.
The impetus for increased robotics installations is global pressure, especially in labor-intensive industries. Lawton notes significant labor content and competitive pressure in China, where he sees heightened interest in collaborative robots. “In China, even with labor costs going up, turnover remains high. Fifteen to 20% turnover per month is not uncommon. Factories there need hot swappable ‘people’ who are immediately trainable. Remember, there is so much cost associated with bringing in new people.”
Collaborative Robots Are Different
You remember the old industrial robots: noisy behemoths bolted to a concrete floor, larger-than-life industrial machines whose arms carried more weight than sensitivity. In some factories, they were industrial accidents waiting to happen. These heavy metal guys were unsafe for humans to be around, and so they were deployed with numerous safety mechanisms like wire cages, red safety buttons and warning signs. Traditional industrial robots were programmed to do a single task, and do it well and consistently every time until a material shortage or maintenance issue stopped them. They had inflexible, limited production applications. Most difficult of all, their software was complex and could only be programmed by experts. Changeover took time, and with manufacturing flows hard-wired through these machines, they maintained a stranglehold on production.
But now, Lawton says, “the days of the industrial behemoth are changing as we bring more intelligence and environmentally sensitive machines to industry. Today, collaborative means that people and robots can work together at the same time. Working without a cage is just one part of it. We can human-scale the task, and establish and run at a human cadence. Collaborative robots can multi-task, and they’ve given us broader definitions of work.”
For decades, manufacturing leaders have sought more flexibility in their operations, often at the cost of quality, high volumes and very competitive delivery times. But the market demands the kind of personalized products that define generations — not just color or fabric choices, but interior options. In the automotive world, transmission, seating, wi-fi systems, even autonomous driving systems must be personalized to sell well. But great variety and traditional robots do not mix well. Traditional caged industrial robots as automation solutions for products with personalized product variety are expensive and it takes time to design a new process around them.
Flash forward to the new generation of robots. Baxter looks more human with his jointed arms and computer screen face. Although Baxter is not a free-range entity, he is cage-free and his feet are not stapled to the floor. And, of course, Baxter never tires. As a general purpose machine he can multi-task over a broader range of jobs.
The Democratization of Robots
While a traditional robot implementation looks like a bucket of parts on a table, a solution that never changes, robots like Rethink’s Sawyer and Baxter have quickly moved in to offer a new vision of factory automation not unlike the progression from room-sized vacuum tube computing machines to palm-sized chips. Think first-generation machines, like Eniac, compared to platforms that run multi-purpose applications like a spreadsheet or Powerpoint on a PC or tablet. Moore’s Law kicked in, generating smaller and cheaper memory, and just as computing machines became general purpose multi-taskers, robotics and factory automation are moving in that same superbly flexible, less expensive direction.
Lawton calls this shift in scale, power, complexity, flexibility and access “the democratization of robots.” He compares the history of computers to what is happening today in robotics. Cray supercomputers, for example, room-sized boxes of blinking circuitry, were tended to by other computers or PhD technicians. They were maintained in climate-controlled, secure facilities, protected from intruders and accidents by heavy-metal locks and sensors. Three decades later, however, we find the same super accessible power in our phones — anyone can buy one, and a single passcode opens the door.
While a traditional robot implementation looks like a bucket of parts on a table, a solution that never changes, robots like Rethink’s Sawyer and Baxter have quickly moved in to offer a new vision of factory automation.
“While industrial robots have been in factories since the 1960s, the truth is that we are only at the tip of the iceberg in terms of what will be possible with collaborative robots” Rethink Robotics founder and CEO Rodney Brooks said. “The world is an extremely variable place, but as software enables robots to operate more like humans, we will see limitless possibilities for robots in manufacturing and beyond.”
Early robots and PLCs (programmable logic controllers) designed by serial entrepreneur Dick Morley (the first of which controlled painting in a GM paint room), could only be maintained and programmed by experts. “But by making machines more accessible,” Lawton says, “we’ve enabled the democratization of robots. When my 13-year-old son visited the lab, in a matter of minutes, he had Baxter deployed. In essence, all the production gear — carts, conveyers, CNC machines — that management needs the robot to interact with can also all be pretty quickly accessed by a human interacting with the robot.”
Old Assumptions, New Robotics
Placement precision still matters, however. “When parts get wheeled up, they must be delivered to precisely the same place every time. So we don’t want to make the parts presentation equipment inflexible either,” Lawton says. The robots that Lawton and his son can use don’t require the same kind of technical grounding that traditional automation required. Lawton believes that although in some cases customers for the new robots may be “guys with long experience,” that kind of experience may not be best for the new robots because “these workers have recipes, all based on the last 40 years, that were used to design the work cell around the robot.”
The same assumptions —that the robot is central to every process and that everything gets bolted down, or that a camera system will be used to line up the same place accuracy every time — no longer apply.
“If we have robots that can feel and move items the way human touch and muscles work,” unwritten rules and assumptions “get in the way,” Lawton says. The shift from the robot as the centerpiece of automated processes, to the robot as an extension of the human, essentially a tool, is under way.
“Very soon, when we hire a production worker, we will provide him with a robot to perform those parts of the job that are unsafe or overwhelmingly repetitive.” Lawton mentions an operation in the Boston suburbs that replaced a human counting cups all day with a robot inserted into the process of stacking and counting. “This approach reminds me of how we did automation in the ’60s when we used a robot every place we could. Everything else was done by a person as a plug-in — only because the robot couldn’t do it!
“Now we can step back and decide whether we need a person, a collaborative robot or traditional automation or robot. The new formula creates happier workers, fewer mistakes, and we’ve lowered the incidence of ergonomic problems.”
When we can make the job better and safer, we know that the combination of robotics and human intelligence excels. “Everybody is really good at something, but asking a person to count cups all day long is like asking a fish to climb a tree.”
Despite the excitement generated by Sawyer and his predecessor Baxter, we know that collaborative robotics is a relatively young field, and that manufacturers must evaluate the costs and competitive benefit of moving to new technologies. I asked Lawton if he could see black holes in the area of robotics applications that must be resolved.
“I see a couple of things,” he said. “The first question I always ask is what the robot should do from a hand perspective. We have done a good job of robotic industrial arms, but hands are so special. We need a much more human-like hand. We’ve researched it for 30 years, and so far a human-like hand is really expensive. There are companies working on this problem like Soft Robotics and Shadow Robot Company, but that’s one of biggest black holes.”
Lawton sees critical manufacturing trade-offs. “For instance if we want the robot to be safe, we have to ask how fast, how heavy. Can it move a 20k payload 3 meters per second at the end of its arm? We may have to make tradeoffs to ensure safety so the control system must still provide a window. If the payload is too heavy, and the example is robots picking up really big tires, then that job won’t be done using a collaborative robot.”
Looking ahead Lawton sees changes that he calls aspirational.
“We’ll have more robots doing things in human ways. We’ll figure out ways to unlock these tasks because there are aspects of work that require judgment, cognition, artificial intelligence, machine learning.” He expects that we will soon start to merge cognitive advances with new types of robots, taking in information, making decisions and applying logic in ways that are more human-like, “although still pretty crude.
“If you get robots that can share and observe and take photos — that are truly collaborative — the robots will learn over time. They will be hybrid brains. They will share information across plants with each other.”