Robots, humans collaborate on safety
In 2008, they were a mere curiosity. In 2012, they were largely viewed as a fad. But just a year later, the industry began to take note. A slew of rivals and suitors entered the fray in 2013.
Now, collaborative robots are here to stay. Many would say they’re the future.
"Collaborative applications is that next new frontier and it’s really going to drive business and applications, and probably applications we don’t even know yet," said Roberta Nelson Shea, global technical compliance officer for Universal Robots, headquartered in Odense, Denmark.
An ABI Research study predicts the collaborative robotics market will surge to $1 billion by 2020, populating manufacturing with more than 40,000 collaborative robots. As that population grows, so does concern over robotics safety.
One of the most anticipated technical specifications in the collaborative robotics realm was released in February 2016. ISO/TS 15066:2016 Robots and Robotic Devices — Collaborative Robots provides data-driven guidelines for designers, integrators and users of human-robot collaborative systems on how to evaluate and mitigate risks.
Nelson Shea is convener for the ISO Technical Committee 299 Working Group 3 (ISO/TC 299 WG3) that was responsible for developing the new technical specification.
Nelson Shea has been involved with the robot safety standards since the first committee meeting in 1982, and as convener for ISO/TC 299 she continues to hold an impartial position in the standards community. She was chair of the ANSI/RIA 15.06 robot safety standards committee for 23 years, and is now chair emeritus. She says the initial idea of collaborative robotics was met by strong skepticism. "The premise about safety was to keep people away from robots. But then the conversation changed to say if the robot with its tool and part touches you and there’s no injury, why not allow contact?"
Factoring in people
"Traditionally, the design of automated systems has not factored in people. But with robots becoming mobile and developing a greater capacity to interact with humans, that design paradigm is not the way of the future," said Roland Menassa, leader of GE’s Global Research Automation Center in Van Buren Township, Mich. "Now I can place a robot with fairly decent capability on the factory floor next to people and they can operate side by side," he says.
Menassa spent 24 years with General Motors. Now he is GE’s resident automation advocate. The Global Research Automation Center focuses on four main areas: robotics, controls, material handling, and work system integration, which tracks the flow of data on the factory floor. GE has embraced the Industrial Internet of Things (IIoT) and automation as a key ingredient. It’s taking the lead on factory optimization, or what it calls Brilliant Manufacturing, to optimize the flow of materials, people and processes within the organization and across its global supply chain.
"When I came to GE, collaborative robots were starting to move on the market, so I visited different factories within GE to do an assessment," said Menassa. "We are either a low-volume manufacturer of very large industrial goods such as gas turbines that weigh thousands of pounds, or human-scale, mid- to high-volume products like lighting fixtures and circuit breakers, where you have hundreds of SKUs on the line.
"We’re still going to weld and have robots handling heavy equipment and performing very difficult processes," Menassa added, "but when you look at where robotics has gone in the last 55 years, we still see a lot of people on the assembly line. And that’s primarily because of the challenges in compliant material. When we make our circuit breakers or lighting fixtures, there are wires and flexible materials that are very hard to handle. The challenge becomes, how do you interject automation in a manual process to handle compliant parts?"
GE utilizes Sawyer, a single-arm collaborative robot by Rethink Robotics, on the assembly line at GE Lighting in Hendersonville, N.C. The collaborative robot inserts components into a LED street light fixture before human coworkers complete the assembly.
"The factory never had a robot for many years," said Menassa. "So to bring a robot inside the factory, we didn’t know what the workers would think. We had a workshop with the factory, looking at the different applications, looking at where it made sense to apply. We then held several campaigns within the plant itself where we actually had the robot on display. We introduced people to the notion of what that robot can do, how you really can touch it, and how you can work with it."
Power and force limited robots are specifically designed to have safe contact with humans by way of inherently safe features of the robot or the control system. These types of robots are typically made from lightweight materials, have force and torque sensing in their joints, and may have soft padded skins.
Four methods of collaborative operation
Under the ANSI/RIA 15.06 and ISO 10218 harmonized robot safety standards and the new TS 15066, there are four methods, or types, of collaborative operation:
- Safety-rated monitored stop
- Hand guiding
- Speed and separation monitoring
- Power and force limiting.
These tend to be the most misunderstood aspects of human-robot collaboration. To avoid confusion, Nelson Shea suggests manufacturers think of each of the four methods of collaborative operation as scenarios rather than distinct modes.
In every instance there is a shared space between a robot and a human operator. In a safety-rated monitored stop, the premise is that in a shared space with a human a robot does not move at all. With hand guiding, a common misconception is that this method is used for teaching. Nelson Shea says that’s not the case.
"When you’re moving the robot’s arm around to teach it certain tasks, this is not hand guiding in the collaborative sense. It’s not running in automatic when you’re doing that," she said.
When used to describe collaborative operation, hand guiding indicates a condition where a robot and a person occupy a shared space and the robot is only moving when it is under direct control of the person.
"In speed and separation monitoring, both the robot and the person can be moving in that space," explained Nelson Shea, "but if the distance between the robot and the person becomes too close, the robot stops, effectively becoming just like the first scenario (safety-rated monitored stop). In power and force limiting, there can be contact between the person and the robot, but the robot is power and force limited and sufficiently padded or otherwise, such that if there’s any impact, there’s no pain and no injury."
She said it’s also possible to have any mix of the four methods of collaborative operation represented in one robot system, even all four of them. The new TS 15066 standard includes formulas for calculating the protective separation distance for speed and separation monitoring. But perhaps the most interesting part of the technical spec is the annex, which contains guidance on how to establish pain threshold limits for various parts of the body, particularly for power and force limiting applications. The data can then be extrapolated to determine speed limits for the collaborative application.
"Although there is information about the four modes of collaborative operation, the more interesting stuff is for power and force limited robots," said Jean-Philippe Jobin, CTO at Robotiq, a manufacturer of adaptive grippers for collaborative robots in Lévis, Quebec, Canada. "More types of these robots are on the market now, but there was no clear guidance except ISO 10218 to help people safely install those robots in their factories."
Jobin co-founded Robotiq in 2008 with president Samuel Bouchard and Vincent Duchaine as a University of Laval spin-off. Jobin is also a technical expert on the ISO committee with Nelson Shea.
Start with a risk assessment
Both Nelson Shea and Jobin stress that the bottom line for any collaborative robot integration is a risk assessment.
"The risk assessment is the most important aspect," said Jobin. "If your application requires a little bit higher force or power than what is stated in the document, it does not mean it is not safe. The data we have from this technical specification is relative to pain, while what is required from ISO 10218 is that no injury should occur.
"There’s a difference between pain and injury," he added. "A user could do tests to show that even if they are a bit above what it states in ISO/TS 15066, it’s still safe because they can prove that the robot cannot hurt or injure the people in those specific circumstances."
Jobin said it’s very important to note that the application is the main concern, not the robot, when assessing risk.
"If you look at the document, it rarely states ‘robot’," he said. "It states collaborative work cell or collaborative application. It involves the cables, jigs, clamps, the robot and the gripper, everything which is inside that cell."
He says it’s a common misconception that if the robot is "inherently safe" then the operation is safe. For instance, if your robot is manipulating sharp objects, then it is not safe to have a human beside it without protective safety measures. Another case is if the robot is handling a heavy object, which could cause injury if it’s dropped or become a projectile at a particular rate of speed.
Safety was a major factor in the robot adoption process at GE Lighting and for instilling worker confidence in the new collaborative robotics paradigm.
"At GE, safety is our overriding priority," said Menassa. "With any application it’s not about if the robot is safe, it’s about ‘is the task safe?’ So we do the task assessment risk-based analysis. We observe all the rules and all the RIA standards. We brought people in from RIA to train us. We make sure we understand what the robot is doing, the shape of the end effector, is there anything sharp, and is there anything that could eventually hurt someone?
"If we feel there is a need for protection beyond just the force or torque limiting capability of the robot, then we’ll place the appropriate safety device, such as a light screen or laser scanner, so we can mitigate the risk," he added.
The ANSI-registered technical report, RIA TR R15.306-2016 Task-based Risk Assessment Methodology, describes one method of risk assessment that complies with requirements of the 2012 R15.06 standard and was updated in 2016.
"RIA publishes the method by which we do a task assessment. We go through the steps of the process and we use their methodology to assess if there’s any risk and how severe it is," Menassa said. "We try to do any of the engineering designs around it to mitigate that risk."
Eyes on the job
Besides its kinematically redundant seven-axis arm, with allows the robot to work in tight spaces on GE’s plant floor, one of the more unique features to the Sawyer robot is its animated eyes on its LCD screen. The eyes don’t actually ‘see’ anything; there are embedded vision sensors in the head and arm for that. The eyes serve a much more human purpose.
"People wonder ‘Why the eyes on the robot?’," Menassa said. "To us, that’s very important. For people to feel safe around these robots, they need to have a sense of what the robot is going to do next. The eyes on that screen are very critical because they look to the left before the robot moves (reaches its arm) to the left."
After GE introduced its workers to the robot, the next step was to introduce Sawyer into the production area. "We took some of the applications that we thought were feasible and mocked them up inside of the plant, but off to the side, where people could start to get comfortable (with the robot)," said Menassa.
"Having a robot do a task is very simple. The challenge in robotic applications is always the material presentation," he added. "Where is the material coming from, how’s it coming, how are we going to grab it, and how are we going to hold it? How are the containers going to recycle? What is the best orientation for Sawyer to pick up and drop off, so you’re not going through the gyrations in 3-D space? That’s a waste of cycle time. If we fix that, the robot can just execute its task."
Value-added from the start
In December 2015, Sawyer joined his human coworkers on the main GE assembly line.
"In any person’s task, whether it’s a 60-second cycle or a 2-minute cycle, you will see 50% to 70% is what we call non-value-added," said Menassa. "It’s when you walk away from the job to grab a tool or a part, to look at a document, or when you’re walking around the cell. When you finally add the part to the assembly, that’s the only value-added time that you spend. The rest of it is non-value. In fact, some of it is waste.
"Sawyer was grabbing parts and putting them in the assembly. But the human was making sure it was fitting properly and inserting the last screws, using what humans are good at—dexterity, perception and logic. For us, elevating the role of the human on the assembly line to focus on quality—to focus on the value-added—is very important.
"It was a tremendous win," he added. "Here’s a business that needs to expand its volume because of demand. The question is how do you achieve that higher output without adding people? Sometimes it’s very hard to add people because of the space constraints. By adding robots, we were able to achieve higher output with the same number of people, at the lowest cost possible with technology that is low cost and flexible. The notion of low-cost, flexible automation is really collaborative robotics."
Menassa said GE also is developing mobile collaborative robots with partner suppliers. "Coupling collaborative with mobility will give us the ultimate vision of having a mobile robot that can navigate the factory and perform multiple tasks, leveraging every minute of that robot, especially in low-volume production. Now we have mobile robots that can tend CNC and additive machines," he said. "Bringing robots to machines will allow GE to stay flexible and agile while synchronizing the scale up of our product volumes with mobile collaborative automation."
Tanya M. Anandan is contributing editor for the Robotic Industries Association (RIA) and Robotics Online. RIA is a not-for-profit trade association dedicated to improving the regional, national and global competitiveness of the North American manufacturing and service sectors through robotics and related automation. This article originally appeared on the RIA website. The RIA is a part of the Association for Advancing Automation (A3). A3 is a CFE Media content partner.