Rethink Robotics has partnered with Shanghai Electric to use a force-sensing robot to fill empty manufacturing jobs in China. Learn why, and how a Roomba founder is involved, in this Q&A.
To help solve China's problem of thousands of unfilled manufacturing jobs, Rethink Robotics has just begun shipping Sawyer, a collaborative robot, to Shanghai Electric, a distribution partner that helps sell robots to manufacturers in China. TechRepublic spoke to Jim Lawton, chief product and marketing officer at Rethink Robotics, about what Sawyer can do, why the China manufacturer is partnering with the small New England company, and how robots are affecting manufacturing jobs in China.
How does Rethink Robotics stand apart from other robotics companies?
Rethink Robotics was founded in 2008 by Rodney Brooks, one of the founders of iRobot that made the Roomba vacuum cleaners. As part of his work at iRobot, he spent a lot of time in China in manufacturing environments. Sixty-five percent of robots sold in the US go into the automotive industry. They weld cars, paint metal, move heavy things around. They do what they do well, but the people missing out on the benefits of using robot technology are, basically, everybody else.
Industries outside of automotive, or small or mid-sized companies, are [often] left out. Traditional industrial robots are very expensive—they take three to five times as much as the cost of the robot to deploy. There's a lot of coding involved. They have to be in a cage, because they're unsafe. If you get in their path, they'll hurt you—they can and have killed people. So in his work in China he's like, "It's odd because there actually isn't very much automation here at all." What he concluded was we really need to be building a very different type of a robot.
The robot that had been designed and implemented up until now has been robots that wouldn't work in the Chinese factories, wouldn't work in industries outside of automotive, and certainly wouldn't work in small and medium businesses. He set out to build a different kind of a robot. We introduced Baxter in 2012. And Sawyer just started shipping this fall.
Some of our largest customers are actually in China [now].
If you're a manufacturer and want to compete globally, you need to be able to do that cheaply. That used to mean manufacturing through low-cost labor-specifically in China.. But labor rates are going up 15% per year in China. It's getting harder to attract talent into a lot of these kinds of jobs. Millennials in particular don't want to be doing manufacturing jobs for the most part. The turnover rate has become really high in many of these jobs. We work with a lot of companies who have 20, 25% per-month turnover. Once a month they're flipping 1/5 of their workforce and they're having to go rehire, so you've got talent-acquisition costs, you've got training costs, you've got re-training costs.
Manufacturers in China as well as the Chinese government said, "Look, if we are going to be competitive on the global stage going forward as we have been in the past, we're not going to do it with labor. We're going to have to do it with technology." China's plan is to become the global manufacturing leader by 2049—the 100-year anniversary of modern China.
They're subsidizing companies that get created to design and build robots. They're subsidizing manufacturers to use robots. All of the major manufacturers in China are trying to figure out how to deploy and design factories and design factory processes that are able to take advantage of robot technology. Programs are helping manufacturers in China build businesses and manufacturing capability that's leveraging robots rather than just leveraging people. At the end of last year, there were 400 robot companies in China—now there are more than 700.
What are the primary tasks Sawyer can do?
Sawyer is doing things like standing in front of a circuit board in circuit testing machines, in front of milling machines, ultrasonic welders. There are tens of thousands of jobs that are literally a person standing in front a piece of equipment. They do something like insert a circuit board into it, push a button, wait for some action to occur like welding or like testing, and then when it's done they take the part out.
It's a lot of people surrounded by testers or by milling machines that are pushing buttons and starting the machine that's actually performing the work. A lot of what Sawyer was designed for, and what it's doing in its applications, are these machine-tending tasks and packaging tasks. At the end of every injection-molded, plastic piece of equipment, you also find a person or two just waiting for the parts to come out.
They grab the part. They might count the part. They might inspect the part, then the put it into a box, so they put it into a bin for further processing. Those kind of tasks Sawyer is doing.
Sawyer and also Baxter use force sensing. What exactly is that is?
The vast majority of robots in the world are position-controlled robots; they know where their endpoint is and then you tell them to go to another end point, then they will go to that point. Their software controls their movement from point A to point B. That's also what makes them unsafe—if you or I happen to be in between point A and point B, we're going to get hit and hurt. In our robots, the outer control will be the force-testing and force-controlled loop. In Baxter's two arms and in Sawyer's one arm, they've got seven degrees of freedom, seven joints.
Each one of those joints has a set of sensors and spring-in devices called series-elastic actuators. They're driving the component of the arms, but they're doing it with this spring-in series. What that allows them to do is flex in much the same way we do as humans. If I push my arm into a table in front of me, my arm will flex. What the force control is on the input and output side of each of these actuators are torque sensors measuring the force and the torque being applied and the torque in the various joints. What that allows the robot to be able to do is feel the forces being applied at every joint in its arm.
By having force sensors on each of the joints within the arms we're able to have these robots perform tasks a lot like the way that a human would perform a task.
At Shanghai Electric, are robots like Sawyer replacing human employees?
As companies are deploying more and more of these robots, they're hiring people that are not roboticists, not automation engineers, but supervisors. Only now, their personnel on the production line are not just people; now they're people and robots. In many cases the people on the line are training. You don't program these robots; you show them what to do.
They're also managing the robot, overseeing the robot in the performance of its task. There are new jobs that didn't exist before that are getting created to manage the new workforce that's partially robots and partially humans. Because the turnover is so high these companies are struggling to fill these kinds of positions. Not a single one of our customers has ever laid anybody off, because what they're trying to do is execute on scale and keep the positions that they need filled, filled.
They're saying, "Let's look at the places in the factory that are jobs and tasks that are really well matched with human capabilities, things like high levels of dexterity, certain levels of judgment, being able to make decision making with abstract forms of data." Those are the jobs that companies are choosing to put their strongest people in. For the jobs that are more built around tasks that are very routine, like standing in front of a machine, pushing a button, waiting for 90 seconds, and then doing other routine tasks, given that they can't hire enough people—those are the tasks that they're targeting the robots to do.
It's not resulting in people being laid off—what it is doing is it's helping them staff their organizations that they're continuing to try to scale.
How is Sawyer constructed?
One of the things that contrasts robots like Sawyer and Baxter with traditional robots is that a traditional robot doesn't really deploy as a robot; it deploys as a project that consists of a robot arm, some PLC, some vision systems, some sensors, some feeders, a conveyor, a bunch of wire. Then the project puts all of that stuff together.
In the case of Sawyer, there is zero customization. Every single Sawyer out there is exactly the same. The only thing that's different from one to the next is you're telling the robots to do a different job. It's as if I went home tonight and sweep up the living room, that's my task. And tomorrow I go home and I clean the garage. I haven't changed, but the task I'm performing changes.
The robots will be told, "Today I want you putting six tubes of toothpaste onto a conveyor to get wrapped up—and that's your job for today, robot. Tomorrow I'll have you standing in front of a testing machine, putting circuit boards into be tested." Those two robots are 100% identical. There's nothing different about the robot or its software.
Could it get retrained to do a different task on a different day?
Yes. What will often end up happening is either they train them to do a different task every day, or multiple times within a day, or they say, "We'll use this robot for these 20 tasks," so when you roll him over to task 19, you say, "Go."
The tasks got there because the person showed the robot how to perform that task and the robot remembered that particular task. He remembers every task he's shown how to do.
It's a lot like when my son was younger and I showed him how to tie his shoes. I reached around behind him, grabbed his hands, and showed him how to tie his shoes. You do the same thing with the robot. You grab its arm, it goes into a new mode. It's the robot using its motor to compensate for its own weight, so it feels like you're moving your hand through free space. You just move his hand where you want it to be. You show him if you want him to grab something or actuate his hand in some form.
Once you've completed showing the robot a task, it remembers that. And this is where the software becomes really important—it thinks in the context of the task. It thinks of what you showed it to do as opposed to a series of Cartesian movements that would normally get programmed in through an interface. The user doesn't have to program anything. It's able to optimize it through its own software.
Did you say there are 700 robot companies right now in China?
Yes, and they added 300 robot companies in the first 90 days of 2015. Manufacturers are seeing this tremendous need—and also, the Chinese government is subsidizing. They're funding the creation of new robot companies. I think it's a testament to how strongly they believe in the robot technology, the role that it's going to play in them being, as a country, competitive globally.
The demographics in China right now are upside down. The population of people needing services as they've aged has drastically outpaced the population of people that are in a position to be able to provide those services.
Even in the US, Rodney Brooks would say he "founded Rethink Robotics to build the kind of technology from a robot point of view that's going to be able to take care of me when I'm old."
The Chinese government has said, "Look, we have clamped down on the number of children and as a result, as they've grown into the age to working age adults, there just aren't enough people." But even if you relax the policy today, you're not going to see an influx of working-age people until they're 18 years old. It's driving them to push really hard figuring out how to build in-country robot companies to help support their expansion plans.
Why is Shanghai Electric using Rethink Robotics rather than one of the 700+ robotics companies in China?
Designing robots is hard. They need to be able to execute the tasks, but they need to be able to do it really inexpensively to be able to get the ROI. The technology to build these robots historically hasn't and still doesn't exist in China. None of the companies that I'm talking about have robots that compete with Sawyer because they can't perform the tasks or they can't perform the tasks cost effectively.
They can't work in environments where people are because, in many cases, these robots are unsafe to be around. They're scrambling from a technology point of view because there just isn't enough of the right kind of robot technology in China to be able to build these companies. It's going to take time to establish that level of expertise. If you think of all of the roboticists coming out of academic institutions like MIT and Stanford and Carnegie Mellon, many have spent time researching and developing the technologies that are required to pull this kind of thing off. That just doesn't exist in China right now.
What's different about robot companies in the US?
It's more consolidated. There are some really big companies like Yaskawa and Fanuc. Their number one customers are in the automotive industry. One of the challenges in being able to perform a task is dexterity. Our hands are actually remarkable, powerful tools in terms of what we can do with them as humans. The research and the advancement in the area of grippers and hands is still a work-in-progress. There are companies like Soft Robotics that has these gel-like fingers that are really good at gripping things that squish, so harvesting fruit and things like that and lettuce, things that you can grip, but you've got to grip in a certain kind of way. A lot of the smaller companies are unique. Robotiq has these highly articulated fingers, hands, that are a lot like our hands. They've got multiple joints in each of the fingers. A lot of the robot companies that are the smaller are those kind of companies.
Is it unusual for Shanghai Electric to go with a small company, like Rethink Robotics, rather than a huge one?
For them it became a very practical question of, "Where are all of these robots going to get deployed?" They had done their own research and looked at "What are the kinds of tasks that are going to need robots?" and then "What are the kinds of robots that are going to be required in order to be able to perform those tasks?" Rethink Robotics is the leader in creating that kind of a collaborative robot. From that point of view we were a very ideal candidate from a partnership perspective. There isn't anybody else that has these mechanically-compliant, force-sensing, collaborative robots. From that point of view, we were the obvious choice.
Where do you see the future of robotics headed?
The technologies that allow these robots to do what they're doing in manufacturing will also be used to create robots that do lots of other things that are outside of manufacturing. At some point in the future, every manufacturer will have this kind of a robot. These kinds of safe, easy to program, inexpensive robots are all going to be in every home. They're going to be doing space exploration and search-and-rescue and elderly care services. The only question is how soon that's going to be. Is it in five years? Ten years? Fifteen years from now?
- Why robots still need us: David A. Mindell debunks theory of complete autonomy (TechRepublic)
- Smart machines are about to run the world: Here's how to prepare (TechRepublic)
- Chinese factory replaces 90% of humans with robots, production soars (TechRepublic)
- Why AI could destroy more jobs than it creates, and how to save them (TechRepublic)