Roberta Nelson Shea, a pioneer of robot safety, chats about how robot safety standards have changed, the worst robot accident she’s seen and what she knows about the much-anticipated UR20 cobot.
At this year’s Automate 2023 conference, the Association for Advancing Automation (A3) has honored Roberta Nelson Shea with the Joseph F. Engelberger Robotics Award, often described as the most prestigious award in robotics. For more than 45 years, Roberta Nelson Shea—currently the Global Technical Compliance Officer at Universal Robots—has been a central figure in the development of industrial robot safety standards in North America and around the world. As the convenor of the ISO committee, she led the introduction of ISO/TS 15066, which is the first document that defines standardized safety requirements for human-robot collaboration.
The Engelberger was previously awarded to another Universal Robots leader, Co-founder and CTO Esben Ostergaard, in 2018.
Engineering.com recently sat down with Nelson Shea to talk about her work and insights in the field of robot safety. Her answers have been lightly edited for clarity.
Engineering.com: What were the original motivations for improvements in robot safety?
Roberta Nelson Shea (RNS): Very early, the motivation was by the big three [GM, Ford and Chrysler (now part of Stellantis)] who said, “If we don’t have a safety standard, labor won’t accept this.” But as time went along the pushback was ‘safety adds dollars.’ Eventually, it was recognized that if robot manufacturers could build more safety in, then it would be easier for integrators, which would make it cheaper for the end users. It took a little bit of time for all this to permeate through the industry. But it was the beginning of the acceptance that robots and robot safety are cost-effective.
Eng.com: Today, safety feels like it’s an expected part of the package and not really top of mind for a lot of people when they’re looking into an investment like this. Do you find that to be the case?
RNS: That’s true with robotics, interestingly enough, it’s not true with other machines. Robots are safer. If you look at statistics or ask OSHA, OSHA says we don’t have a problem with robots, because people end up safeguarding the robot. But you will find—and it is the end user who determines this—there are end users that are very conscious of safety and they really want to do the right thing. And for every one of them, you have others who are not as engaged or just not really understanding it.
Eng.com: What was the worst accident you’ve seen?
RNS: This was before working for UR; I sometimes did safety consulting. I did one accident investigation on a press where a worker lost both hands. I did the machine review and the machine failed. The company knew it needed a complete overhaul of the control system but didn’t budget for it. They got a quote for it five years previous, and they decided that it cost more than they wanted to spend. They got the quote updated every year, as if to think about doing it, and never did it.
Eng.com: What has changed since then?
RNS: Generally speaking, robotics is more recent. We seem to have gotten people engaged and willing to move forward on safety. But we had to get over a hump in which there would be people saying: ‘But why do I have to do this?’
I think A3 helped out a lot with this back when they were RIA [the Robotics Industry Association], in that they put on a lot of training about risk assessment. We talked about a lot of application stories. It’s an industry that has quite openly discussed what hasn’t worked well. You will not find that really in other industries.
Eng.com: Can you give me an example of something that hasn’t worked out well and how it was amended?
RNS: Well, there was a very early fatality, I think it was in Jackson, Mich., in which a person got in through the perimeter guard because there was an opening that was only 24 inches. I can’t get through that. And he went inside to just observe something because he was seeing some glitch. Frankly, most of the people around robotics that do get injured—and there are not many—but they tend to be the people who are engaged and really want production to work well. So, the worker went in to have a better look. And the robot went through its normal motion, but where he was standing meant that the backend of the robot hit him in the chest as it swiveled. Usually that would not have killed him. But he stood in front of a post that was installed to prevent the robot from exceeding its defined work area.
Ironically, the thing they installed to prevent the robot from crossing that position ended up being a crushing point. The discussion was not about what was done wrong there, because you can readily see that somebody identified, “Hey, we want to make sure that the robot doesn’t bash through the fence.” And this thing that was a safety post ended up introducing another hazard which was not fully recognized because the thought was nobody would ever be inside the perimeter guard.
What we really had to understand is why does somebody want to get in, and can we eliminate that need? That is why we’ve really pushed hard on root cause. Not only that, for the reason why that post was in there, there was also a big cry out from the safety committee, R15.06, to the whole robotics world, saying, hey, we need to have ideas about how we can better ensure that a robot won’t be able to exceed its motion limits without introducing hazards.
Eng.com: How can a smaller company have the bandwidth to understand and focus on safety requirements?
RNS: That’s a challenge I work with on a daily basis. As soon as you get out the very large players with strong unions and strong safety cultures into smaller places where—just because of the numbers of people—the odds of things happening are much lower. For these smaller ones, they don’t tend to have the expertise. I can tell you that there are some small enterprises that buy robots thinking ‘now they’re so smart, people aren’t going to be hurt.’ And that’s something we grapple with. UR came into the world with very small robots. Let’s be realistic, the risks associated with a three-kilogram payload are very different from the risks associated with a 100-kilogram payload. And by far the majority of robots by units in the field are over 100 kilograms.
Eng.com: Cobots have made automated welding, material handling, sanding and finishing more affordable than ever before. What should SMBs keep in mind on how to approach safety with all this new available technology?
RNS: The reality is, there’s no such thing as a safe robot. It’s the specific application that must be made safe. Even a tiny robot—if you equip it in an application where there are toxins, a torch or a knife—can be unsafe.
For a small enterprise, be vigilant about who you choose to do business with, ask a lot of questions. The most important thing to ask is “Do I think that the robot I’m buying has a name or a brand that is just not going to fold up in the night?” That’s just common sense.
For an integrator, [robotics system] OEM or robot-as-a-service (RaaS) company, are they a credible supplier? Ask about their risk assessment and if they would review it with you. I also encourage everybody to ask, “How will my people be trained to safely handle this? Do I need to add any safeguarding? The supplier won’t know what your plant floor looks like—there’s a lot of things that the supplier may not know—there should be a conversation and a willingness to discuss it.
Eng.com: Are there any recent trends in robot safety that you are hesitant about?
RNS: If anything, I’m a little bit dismayed that people want to make everything cage free—I don’t understand it. People say it’s because [when it’s caged] you can’t move the robot around. But you can mount guards to the cart! I just look at it as a very commonsense thing.
Companies that mount robots on a cart to be transportable, they kind of fall into two camps. One ends up finding the place where they can best use it and never move it. The other one is more or less buying an OEM solution with safety built in. In fact, I just saw a really nice example of this for palletizing at Automate 2023. It was a solution for a palletizer, which includes area scanners. It’s installed together and it unbolts into two pieces so a forklift can move it wherever you want and then you bolt it together. That does make some sense for palletizing, particularly in some older factories where they built a line to make some goods, and then they built another line over there to make some goods, but they don’t run them all the time. Do you want to redo all conveyors and your whole factory floor, or do you want to every now and then use a forklift to pick up and move these two pieces and then use it at another place? And I thought that was quite clever—something made to be modular for not just shipment to the customer, but for the customer to move around.
Eng.com: UR is releasing the UR20 soon, which will be a bigger cobot. Will it need strong safety protocols?
RNS: That’s correct. The UR20 has the same safety system that we have in our e-Series, but the UR20 has all new joints. It’s a beautiful design where the joints have 50 percent fewer parts. People will get a cobot with the reach, payload capacity and speed of a traditional 20-kg to 25-kg payload robot, so The UR20 is going to have to be treated more like an older industrial robot. You still get all the safety features, of which most are very useful. However, you don’t really find great value from force limiting to protect a person, because you want to run it fast with high payload—that’s what it’s for. People should use a robot to their best advantage. Use the safety features to lock down how far the movement is, the directions of the movement, or decrease speeds if needed.
Eng.com: What is the most important, influential element of robot safety that has been written into the standards over the last 20 years?
RNS: Our emphasis on task-based risk assessment. Before that, if you were a safety professional, you would identify hazards visually—just looking for obvious things. With task-based risk assessment you ask “What are all of the tasks associated with the setup, operation, troubleshooting and maintenance of the robot and the area where it’s used.” And then you ask “What are all the hazards an operator is exposed to?”
It sounds really simplistic. You just write a list of tasks, and you separately have a list of all the hazards, and then you match up the task with each of its associated hazards. Do we have an electrical hazard, a trip hazard, high pressure, clamps, impact by robot, impact by end-effector? Task-based risk assessment was something of an ‘aha moment’ that came about in the 90s in robotics and it has filtered out to just about every use in safety.
Eng.com: What is your final piece of advice for our readers when it comes to robot safety?
RNS: Focus on the root cause, use task-based risk assessment and fully engage people. The more people involved as a part of the risk assessment, the more people that view it as their “baby.” And these are the most important things you can say to people about robot safety: Engage all ideas, all opinions are of value.
Eng.com: I think that goes throughout life, doesn’t it?
RNS: That’s right.