Collaborative robots

Feb. 23, 2015 – Fifteen years from now, there will be a collaborative robot in every single manufacturing environment in the world. For sure. That prediction comes from Jim Lawton, chief marketing officer of Rethink Robotics, a Boston, Mass.-based robotics company.

Rethink is one of several companies with collaborative robot offerings on the market, helping to bring attention to this new category of robots.

The term collaborative robots refers to humans and robots working together on common tasks, safely. It’s become an industry buzzword, and is even addressed in the latest version of the ISO 10218 — robots and robotic devices — standard, where collaborative operation is defined as “a state in which purposely designed robots work in direct co-operation with a human within a defined workspace.”

But there are certain applications better suited for the emerging technology, and certain considerations need to be made when deciding whether collaborative robots are right for your manufacturing environment.

What are the opportunities?
Collaborative robots emerged on the manufacturing scene because the industry was demanding flexible automation to replace dangerous, boring and error-prone tasks at an affordable price, and many manufacturers were looking for alternatives to chasing low-cost labour across the globe. At the same time, robot manufacturers were looking for ways to give customers the benefit of more human collaboration with robots.

Today, the market is growing, as robotic companies work to build safe, collaborative robots to help customers take advantage of the opportunities the technology offers.

But in order to understand the opportunities, you must first consider what collaboration looks like. And that depends on who you ask.

Erik Nieves, CTO at Yaskawa Motoman, says that collaboration technically means four flavours — stopped state monitoring, speed and separation monitoring, hand guiding, and power and force limiting. The first three, he says, are supported by just about every robot controller out there. However, it’s the latter that’s typically referred to when collaborative robots are discussed. But Nieves says that’s too narrow.

“When we first started thinking about the notion of collaborative robots, it was simply about having robots and their human operators work more closely and naturally together. So it really was just that. How do I make this system more efficient by leveraging the strength of the robot and the strength of the person in better balance?” explains Nieves. “So the robot is good for strength and memory. They can repeat tasks and they can do inhuman tasks in terms of payload and repeatability. But the human operators are much better at decision-making and quality control. So we view collaboration as being very broad, as having direct applicability to lots of robot applications.”

Rethink’s Lawton says collaborative robots are ideal for applications that require flexibility and versatility. For example, a manufacturing environment with a high-mix, low-volume operation would be a good candidate, as well as applications “where the task needs to get done in very human-like ways.”

Packaging environments are a good fit, as well as applications that require line loading and unloading, machine tending, material handling, light, general assembly, testing and sorting. Applications where collaborative robots may not be the best fit are those that require extraordinary precision.

Nieves believes collaboration has a role to play in process robots — those adding value, whether it’s painting, welding or another process. Material handling applications, where robots are supporting moving parts in and out of production, are another area where he feels collaboration is useful — even in applications requiring material handling over 10 pounds. Another application is material handling less than 10 pounds, where precision is not required.

Tom Moolayil, technical support manager, North America, for Universal Robots (UR), says the biggest application in North America for the company’s UR5 and UR10 flexible, lightweight, six-axis robot arms is CNC machine tool tending.

With these new, portable five- and 10-kg robots, you don’t need to worry about floor space, which is a huge bonus, says Moolayil, “because you don’t have to worry about guarding [and] you don’t have to worry about concreting the floor and bolting it down because it’s a lightweight robot. So you can make it do one thing here, you can move it around, program it to do something else. And that makes it a lot more flexible to install this robot in [a] plant.”

But does collaboration automatically mean the robot is not fenced? Not necessarily, says Nieves, giving welding as an example. You can gain the benefit of collaboration in loading the part directly without having this interim fixture in between, but it’s still fenced, he says.

“Why? Because the robot needed to shield the operator from the arc flash. So there is a zone where the robot comes out and presents itself to the operator in the wide open. But when it’s going over to do the work, it’s hiding itself in the cubby where the arc welding takes place. Not all fences are about keeping people away from robots. There are reasons for barriers that go outside of human-robot collision,” he says. For example, process paramaters, hygiene and throughput.

But there are a growing number of applications where you can legitimately eliminate fences. It all depends on the application and what the robot is holding, says Jim VanKessel, owner of JVK Industrial Automation Inc., in Kitchener, Ont.

“The big trick with collaborative is, what are you putting in the hands of that robot — the type of grippers, the type of product you’re handling? Some will require guarding of some sort, like light curtains, where other applications don’t. So the real trick is assessing the risk of the specific application,” says VanKessel.

“The robot by itself has all the safety built into it. The part that’s important is assessing the risk of the end-of-arm tooling, the gripper mechanisms — whatever they’re using to handle their materials. It’s assessing that risk of the specific application…The robot by itself is fine, but you put a sword in it, it’s every bit as dangerous as a person with a sword.”

What makes them safe?
Rethink’s Baxter robot has a four-kg payload and features series elastic actuators. This technology uses springs to advance the robot’s motion control solution from one of rigid positioning to one of force control. This, according to the company, gives Baxter its smoothness of handling, so it won’t damage parts, fixtures and materials, or harm humans during operation. In addition, its smooth, pinch-proof exterior, combined with back-drivable motors and force-detecting sensors, are designed to minimize the impact of inadvertent contact. Embedded sensors allow Baxter to “feel” and compensate for common task variables in order to place parts into a fixture or alignment. The robot recognizes and adjusts to subtle changes similar to its human co-workers, because Rethink designed it to operate in “human-like” ways.

“There are four sensors in each of [Baxter’s] joints,” says Lawton. “Any aspect of that arm can come in contact with you, and will immediately be able to respond and understand that it’s experiencing a force and the force can be controlled in a way to prevent somebody from being harmed by that. Because the springs are in series, there’s also a certain amount of give to the arm that absorbs some of the energy, so that when you do come in contact with it, the springs are absorbing some of the impact of that rather than you.”

Lawton says Baxter is so safe, none of its customers have it caged.

“We don’t have a single instance anywhere of a robot being in a cage or being guarded in any way. There’s not one that’s being guarded,” he says.

“There is an envelope that collaborative robots are going to have to operate in to be able to maintain safety,” explains Lawton. “There are some tasks that require heavier payloads, but when you start to combine a really long reach with a heavy payload, with a lot of speed at the tip, that’s when you get all the kinetic energy that can actually hurt people…And in those circumstances, that’s what’s going to require the thing to be caged.”

Universal Robots’ UR5 and UR10 robot arms feature “smart joints,” says Moolayil.

“They have force sensing built into these joints. It’s continuously monitoring the force on each joint. And if there’s a force spike sensed in any of these (i.e., if it runs into something, there’s a force spike)…it will just stop. That’s what makes these guys safe to work with,” says Moolayil.

UR’s third-generation robot arms can operate in reduced mode when someone enters the robot’s work cell, and then resume full speed when the operator leaves. Or the robot can run full speed inside a CNC machine, for example, and reduced speed when outside. The company’s patented safety system monitors eight safety functions: joint position and speed, TCP position, orientation, speed and force, as well as the momentum and power of the robot.
For Yaskawa’s offerings, it’s the Functional Safety Unit that makes most of its robots (equipped with its latest controllers) safe to use in collaborative environments.

“It’s the functional safety. Collaborative means, from a controls perspective, it means you can be in the safeguarded space of that robot. You’re within reach of that robot, and you can be there safely,” Nieves says.

With these and other collaborative offerings on the market, VanKessel says a risk assessment is necessary to make sure the application is really safe.

“I’ve had a couple of people now try and say, ‘We don’t need guarding because it’s a collaborative robot.’ But at the same time, they’ve got a pin on the end-of-arm tooling that’s an inch and a half long and three eighths of an inch in diameter. Well if a three-eighths-of-an-inch pin hit me in the head, it’s going to do some serious damage, even if it’s only five pounds or a five-kilo application. It’s going to crack my skull. And it’s assessing those risks that drives the whole project,” VanKessel says.

It’s going to be a learning process for manufacturers, says Esben Ostergaard, founder and CTO of Universal Robots.

“You can’t just put a robot up without fences and say it’s a safe robot, because you really have to consider the risk in the installation,” he says.
Though the Danish company says that, after initial risk assessment, 80 per cent of its robots are used in collaborative environments without safety guarding.

Are collaborative robots right for you?
Customers are embracing this new class of robots, says Nieves, because it allows them to be both safe and productive.

“Their productivity goes up, for one, and they haven’t traded productivity for safety — they’ve just gained productivity.”

But there are several factors to consider before bringing collaborative robots into your environment. Lawton says it really comes down to whether the robot can perform the task, and whether it can do the task cost effectively.

“Think broadly about what one means when talking about collaborative robots. It’s not just about getting them out of the cage. What you really need them to do is perform tasks and have very human-like characteristics in the way they approach solving problems. When you think about them through that lens, it helps you identify the best places to deploy them [and] how to make them successful,” says Lawton.

But for VanKessel, it all comes down to that risk level.

“If the application is applying a lubricant to a product, it’s very easy to get a lubrication dispenser on the end of the robot that’s very safe. What’s the impact on the people around it? If we take a robot and ask it to do material handling, what does the part look like? Does it have sharp edges? The minute it has a sharp edge, it could hit me in the head, so now we need guarding.”

Nieves says to look at your application, your use case.

“Your use case will point you in the direction of collaboration or not. And if collaboration, which flavour of collaboration makes the most sense for this particular application and installation?”
And when you’re at that point, says Nieves, the first thing is to look at the mechanical performance.

“What do you need to pick up? What do you need to handle? What do you need to do? So you still have to think in terms of payload and reach.”

Next, he says, consider what the operation looks like. Is the robot by itself? What is the level of interaction? What does collaboration look like? How portable do you need the system to be? And what other hazards does the application present?

What does the future hold?
The industry is still at the front end of this emerging market, but Lawton predicts a lot of innovation ahead. “There’s a lot more excitement ahead in terms of how all this comes together over time. There are just so many really interesting things you can do with collaborative robots,” says Lawton. “Over time, we’re also going to see that these robots are going to be able to be leveraged in ways that harness the power of big data and analytics, and the ability to learn from their experiences.”

He suggests that, in the future, robots will be able to learn from their experiences and change their own behaviours, and communicate these changed behaviours to other robots they’re working with.

This is just the beginning.

“There’s no question that collaborative robots are here to stay,” says Lawton. “Every manufacturer is going to have a collaborative robot. In fact, many of them are going to have thousands of collaborative robots. And people won’t be competitive if they don’t.”

Glossary
There are four “flavours” of collaboration that Yaskawa’s Erik Nieves describes.
• Stopped state monitoring – The robot stops when a human enters a scanned area, but continues to monitor the area until the operator leaves, at which time it resumes working.
• Speed and separation monitoring – The robot slows down when a human comes near, and may stop if the person gets too close.
• Hand guiding – The user is in direct contact with the robot while guiding and training it.
• Power and force limiting – Safety is achieved through restricting the amount of force available in the system.
These robots are limited in payload, acceleration and torque.

This article originally appeared in the November/December 2014 issue of Manufacturing AUTOMATION.