Industrial Mobile Robot Safety Standards on the Forefront
| By: Tanya M. Anandan, Contributing Editor
Mobile robots are entering new territory usually reserved for humans and manned vehicles. In warehouses and distribution centers, to factory floors, and now laboratories and office spaces, we can expect to see these free-ranging robots in more scale. Get a load of these warehouse wranglers. These are more than just mobile pack mules. These robots have on-board intelligence and real-time adaptive capabilities that broaden their approach.
A year flew by since we published our primer, Mobile Robots and Intralogistics the Always-On Supply Chain. The market is now stacked with more choices, combinations and applications. A plethora of variations that pose new challenges for safety standards writers racing to keep up. Many of these mobile platforms are already in practice or in pilot programs around the world.
Check out this video of a dual-arm mobile robot on the job at weld assembly equipment supplier HIROTEC AMERICA. In this application, an OTTO 1500 self-guided mobile platform is coupled with a Yaskawa dual-arm robot (pictured) to automate the monotonous task of dipping parts into black oxide for corrosion protection. Robotiq makes the adaptive gripper you see deftly handling the parts as they are lowered into the black oxide.
There is no safety fencing around this spry robot. That’s because OTTO uses safety-rated LIDAR sensors to detect and navigate around obstacles within its path. According to the manufacturer, OTTO 1500 has two safety-rated LIDARs, one on the front and one on the back of the platform. It has four emergency stop (E-stop) buttons for manual emergency stopping. This mobile manipulator helps HIROTEC move closer to its goal of lights-out manufacturing for 24/7 operation.
Autonomous, Self-Guided Vehicles
Consumer goods giant Procter & Gamble is also putting mobile robots to work in multiple sectors and varied applications.
“Because P&G is such a diverse broad-based company, we have many different types of mobile platforms in use,” says Mark Lewandowski, Robotics Network Innovation Leader at Procter & Gamble in Cincinnati, Ohio. “All the way from large-scale mobile platforms, to high-lift mobile platforms, and all the way down to small mobile platforms that can be used in office or lab environments where there are more people. Between manufacturing, warehousing and distribution, we have many different needs across a broad scale of sizes, shapes and platforms.”
Lewandowski notes that these are all self-guided vehicles. This means they navigate autonomously to determine the best route from point A to point B according to pre-learned maps of their environment. Using a variety of sensors and software, they navigate around unexpected obstacles with real-time intelligence.
This video shows an example of how these types of mobile robots achieve mapping and route planning for easy deployment. This is in contrast to a traditional automatic guided vehicle (AGV) that typically requires existing infrastructure or facility modifications, such as magnetic strips or navigational beacons embedded in the floor, to guide the vehicle on a designated path.
Not Your Grandfather’s AGV
P&G is either using or testing three main categories of industrial mobile robots: large mobile platforms with high-payload capacities, smaller self-guided vehicles, and mobile manipulators (mobile robots combined with a robotic arm).
“The biggest and most active (application of mobile robots) right now is in the realm of autonomous mobile platforms,” says Lewandowski. “These are large-scale platforms that move around pallets or large unit loads. The mobile platforms are in our warehouses and distribution centers where there is not a lot of human and vehicle interaction.”
He says they use smaller autonomous vehicles for moving input materials, work-in-process, or finished goods in between processes. These smaller platforms are able to move around more nimbly in crowded environments.
“An example of an application we’re piloting right now is moving quality and lab samples from production processes to the lab where they need to be tested,” explains Lewandowski. “This involves bringing these smaller vehicles into an environment with a lot people and changing processes, so the vehicles need to be able to adapt to these highly dynamic environments.”
P&G is already using or developing pilot applications for mobile robots such as Clearpath’s OTTO, Aethon’s TUG, and Omron’s Autonomous Intelligent Vehicle, as well as autonomous mobile platforms from startups like Fetch Robotics, Mobile Industrial Robots (MiR), Canvas Technology, and Milvus Robotics.
“Safety is fundamental,” says Lewandowski. “Many of these mobile systems are based on the same technology (LIDAR), so from a safety standpoint, the point of entry is to make sure you have what is recognized as a capable safety system to detect objects and people and to react appropriately.
“Some of these new people (startups) coming in are starting to use these vision-based types of technologies that have more spatial awareness,” he continues. “It’s important to not only know where the platform is going on the floor level, but also know where it’s navigating in its environment. Are there doors or overhangs, or tables or conveyors? It needs to be able to see those and react appropriately to avoid those types of obstacles as well.”
Unlike their fixed platform siblings or even their cousins on the traditional AGV side, autonomous mobile robots are free-roaming. Safety is a major concern.
Safety Standard in the Works
The Robotic Industries Association (RIA) in conjunction with the American National Standards Institute (ANSI) is developing safety standards for industrial mobile robots. RIA has assembled the R15.08 committee to write the standards documents for Industrial Mobile Robot Safety.
Work is already underway, with the first two parts of the standard addressing safety guidelines for robot manufacturers and integrators expected to be drafted by the end of 2017. Then the document will have to go through a series of review processes and any revisions if necessary. The goal for publication is early 2019.
Why so far away you might ask? Actually, this standards development process is moving at an accelerated pace.
“I’ve been involved with the safety standards (process) since 2007. To me, this schedule has been lightning fast,” says Michael Gerstenberger, PE, a roboticist and the Chair of the R15.08 committee. “There are several companies coming out with products in this area. RIA wants to get on top of it and make sure we serve the membership well by developing pertinent standards for how to do things safely.”
Gerstenberger has an interesting challenge with this committee. R15.08 is tasked with bringing together two worlds, the mobile platform community and the industrial robot community. Each is represented by its own safety standards.
For mobile platforms, the closest existing standard is ANSI/ITSDF B56.5-2012, Safety Standard for Driverless, Automatic Guided Industrial Vehicles and Automated Functions of Manned Industrial Vehicles.
For industrial robots, the standard is ANSI/RIA R15.06-2012, Industrial Robots and Robot Systems – Safety Requirements.
“Many of the new mobile robots, however, possess a greater degree of autonomy than the traditional AGV, enabling them to operate safely in a more dynamic environment where real-time path re-planning is crucial,” says Carole Franklin, RIA Director of Standards Development in Ann Arbor, Michigan. “This is not an area currently addressed in the existing AGV standard (B56.5).”
She continues to explain that while the industrial robot safety standard (R15.06) allows for the possibility of a mobile robot system in an industrial environment, most of its guidance assumes a fixed, stationary robotic installation.
Together, committee members must develop this new ANSI/RIA R15.08 standard that will bridge any gaps between the two existing standards, B56.5 and R15.06, and provide guidelines for safety considerations unique to mobile robots. In the end, they may decide to fold the newly assembled guidelines into an existing standard, but that remains to be seen.
“One of the most important things that I’ve been trying to do is to make sure that all the voices on the committee are heard,” says Gerstenberger. “I want to make sure that everyone can express their opinions. We have pretty good representation on the committee.”
The 75-plus committee members represent most of the traditional robot manufacturers, established mobile robot suppliers and startups, collaborative robot suppliers, traditional AGV manufacturers, robot integrators, safety systems suppliers and integrators, and major end users in the automotive, electronics, consumer goods, and logistics industries.
Also represented are key industry standards and safety organizations such as TÜV Rheinland of North America, Underwriters Laboratories, the National Institute for Occupational Safety and Health (NIOSH), and the National Institute of Standards and Technology (NIST).
Gaps in Existing Standards
NIST develops test methods and metrics to evaluate the performance of new technologies. Occasionally NIST will lead the efforts toward standards development, as they did in this case with R15.08. For several years, NIST has conducted research in its labs on traditional AGVs, mobile robots, and collaborative robots, and the hybrid integration of these machines.
“We’re interested in taking robot models from all different manufacturers and integrating them into one workcell,” says Jeremy Marvel, a computer scientist at NIST and member of the R15.08 committee. “That also encompasses mobile industrial robots, particularly the ad hoc integration of the robot arm with mobile platforms from different manufacturers.”
Marvel and NIST colleague Roger Bostelman, who is also on the R15.08 committee, studied the existing safety standards for AGVs and industrial robots while they were conducting their research on hybrid systems. In some cases they found sufficient overlaps, but in other cases they found gaps and even conflicts. You can check out the results of NIST’s research in this paper.
Gerstenberger says the committee continues to use NIST’s findings as a basis for making sure they have the gaps and conflicts covered as they draft the R15.08 standard. He notes one of the conflicts.
“In the R15.06 standard (published in 2012), it says the robot can be fixed in place or mobile, but shortly after that it also says that the standard only covers the manipulating portion of the mobile robot. So it doesn’t cover when this robot can run around, when it’s no longer caged in an area.”
He also addresses one of the gaps.
“How do the controllers communicate safety information? Should an emergency stop button that’s on the mobile platform also cause an emergency stop action on the arm? What about the other way around? If there’s an emergency stop that’s clearly part of the robot controller, should it stop the mobile platform?”
The R15.08 committee will have to tackle these questions and many others as they draft the document.
“That was the basis for R15.08,” says Marvel. “We want to be able to address those gaps and conflicts, and provide guidance and requirements for the provision of a safety-rated solution.”
He notes an example with teach pendants: “If you’re considering an ad hoc installation or integration of a mobile platform with an industrial arm, both of these will have their own teach pendants. How do you rectify or remedy the control issues so that you only have a single point of control? Do we specify a way to accomplish that? Do we specify that the mobile platform has to relinquish control to the industrial robot, or vice versa? Or do we specify generically that there be a single point of control and then let the integrator or manufacturer figure out how to do that?”
Marvel says right now the committee is leaning toward the latter, meaning it will be up to the manufacturer to determine the single point of control when adding a manipulator to its platform. It’s up to the integrator when integrating the two together.
“It’s a significant challenge to integrate two disparate manufacturers’ systems,” says Marvel. “In most cases, these are closed systems (with proprietary technology). These are not systems that are meant to be controlled with someone else’s software.”
Marvel says it’s still not easy, but easier to use a common open language like the ROS operating system.
“It took us several months to do our first integration with a closed AGV system, and we’re still not sure if we did it correctly. Versus less than three weeks for an open platform we just integrated this summer.”
NIST took a conventional forklift AGV and mounted a Universal Robots UR10 collaborative robot arm on top of the mobile platform, making it into a hybrid (pictured).
“The hybrid does not use guided wires, magnets, or line-following (like traditional AGVs),” explains Marvel. “Instead it uses a laser-based triangulation system to figure out where it is within an open environment, allowing the control software to tell it where to go and generate paths dynamically. Once it’s been told where to go it will not deviate from that path. We’re working with the manufacturer of the AGV platform as well as the control software to make it more agile and more adaptive, so it can respond in safe ways to any potential collisions or safety hazards.”
They also have an Omron mobile platform, which is compatible with ROS. They are comparing closed systems with open systems.
“It’s really hard to get safety signals integrated. Same thing with any kind of control. The robot controller is used to being in charge of everything. So now when you’re trying to integrate two intelligent systems, it becomes a little bit of a negotiation. Trying to negotiate with two systems that don’t want to negotiate is tricky. You end up with a left hand-right hand situation, where the robot arm and the mobile robot are doing their own things, but they are more or less oblivious to each other. That intersection starts to get a little dicey in terms of safety and reliability.”
Mobile Robots, a Moving Target
NIST’s research sheds light on the complexity of developing a safety standard for industrial mobile robots. Devising a safety standard for an autonomous mobile platform is one thing, but now add a robotic arm or some other type of attachment, whether it’s shelving, a conveyor, or tote, or even another vehicle towed behind it, now the safety implications become more intricate.
We’ve seen these types of systems out in the wild. Take the previous application at HIROTEC, and now this Bastian Solutions mobile platform integrated with a Yaskawa arm (pictured) for robotic batch picking. Equipped with a vision system, Bastian’s mobile robot serves as a robot-to-goods solution by picking multiple types of products, eaches, and SKUs for various orders at a time. See it in action.
According to Bastian, this dynamic mobile robot uses safety scanners to detect and avoid both obstacles and people in its workspace. It has a Class 1 E-stop for emergency shutdown. The mobile platform adheres to several safety standards including ANSI/ITSDF B56.5 for driverless, automatic guided industrial vehicles.
“Ultimately, if you can get the mobile platform and the robot arm to move in tandem, and able to coordinate their motions such that you can place a tool specifically with high repeatability and accuracy anywhere within your virtually limitless work volume, that right there is a fantastic enabling technology,” says Marvel, envisioning the possibilities.
Think big. Airplane fuselages, ships, trains, even buildings. Check out SAM, the mobile construction robot.
“This also enables a highly dynamic work environment, one in which your assembly line can reconfigure itself to accommodate new processes or new batch jobs,” says Marvel. “Goes with that agile manufacturing paradigm.”
As industry moves ahead with mobile robot pilot programs and more adoptions, the R15.08 committee will strive to keep up. Marvel notes more challenges ahead.
“For instance, if you have a mobile platform that for some reason shuts down or goes into an emergency stop, the attached arm needs to know this. It may actually need to know why, especially if it’s trying to do a dynamic control situation where the motion of the mobile platform and the arm is coordinated to keep a tool in a specific Cartesian coordinate.”
He says the same challenge arises for control signals.
“The robot will tell you where it thinks it is. The mobile platform will tell you where it thinks it is. These signals then need to be integrated together. In some cases, we find that reported values for position and orientation aren’t always what is internally represented.”
In other words, by the time you get the report back from the robot, it’s not in real time. That’s old information.
“There could be delay, jitter or lag, or just uncertainty about the positions as reported that can dramatically impact performance and safety of the entire integrated solution,” says Marvel.
The R15.08 committee will need to take all of these considerations into account while writing the new safety guidelines. End-user input is critical to the process.
End-User Insights Vital
P&G’s Lewandowski and colleague Bob Bollinger, Beauty Sector Robotics Leader, are both on the committee along with many other end-user representatives from a range of industrial sectors. They say it’s important to “check your company logo at the door” and discuss what’s best for the technology as a whole.
Lewandowski has over 25 years with P&G. Bollinger has over 30 years in industry, 18 with P&G. Both are part of the company’s Robotics Technology Network. The RTN is a group of P&G employees that leverages its collective experience with robotic technology to determine how best to deploy the robots within the company. The network also evaluates and tests new robot technologies.
“As an end user, we need to be sure the standards are clear, usable and deployable, and that the products produced as a result of these standards can be used safely in the plants,” says Bollinger. “I think end-user participation really helps ground the committee and determine what the use cases and challenges are. Hopefully we can come up with clear guidance on how to approach those challenges.”
One of the first challenges the committee faced was to determine the extent of its scope. In an industry sector that is still evolving, with new startups and technologies entering the arena every few months, it’s been difficult to narrow the focus. After a couple of committee meetings, RIA’s Standards Development Director says they were able to nail it down.
Industrial, as it applies to the scope of the R15.08 Industrial Mobile Robot Safety committee, includes only the industrial use of mobile robots in the production of goods and related services. This includes structured and semi-structured environments in manufacturing, warehousing, and the logistics space.
“It is expected that people (adults only) within proximity to mobile robots in the industrial space will have some level of training, so they know how to interact properly with the robots,” says Franklin.
Outside of the scope are mobile robots that have interaction with the public, such as hospitality robots, personal care bots, healthcare and rehabilitation robotics, and robotic amusement rides. The guidelines will not cover these applications.
Mobile in this context refers to only ground-based systems. Airborne and waterborne systems are out of the scope.
Also out of the scope are platforms that are ridden, such as manned forklifts. Remote-controlled or tele-operated mobile systems are not included. Mobile systems on rails, such as gantry robots, are also beyond the scope.
Robot includes the mobile platform itself, even without a robot arm mounted on top. But the mobile robot must have autonomous capabilities beyond a traditional AGV. Traditional AGVs will remain under the Industrial Truck standard (B56.5).
Franklin says the R15.08 document will be silent on the topic of modes of mobility, whether wheeled, uni-ball, tracked, or legged platforms. That means Boston Dynamics’ bipedal robot is a candidate. That is if Atlas is stacking boxes in the warehouse, assuming this mobile manipulator can maintain its footing.
The safety standard will provide guidance to robot manufacturers for the mobile platform, the attachments (shelves, conveyor, or robot arm), and an integrated solution (mobile platform + robot arm designed and manufactured as single unit with one controller). For integrators, guidance will be provided for integrating a mobile platform with a robot arm.
Guidelines will also include hazard identification and risk assessment, plus address lockout/tagout (LOTO) recommendations for control of hazardous energy.
Part 1 for manufacturers and Part 2 for integrators is expected to be drafted by the end of the year. Part 3, which will provide guidance for end users, is deferred until a later time.
Charting New Technology
An important part of the standards writing process is anticipating technology advancements, especially when the standard will help shape technological maturation in a developing area like mobile robotics.
“Technology tends to change at a more rapid pace than standards can be updated and developed,” says Lewandowski. “The ultimate goal of the standards maker is to try to anticipate where things are going and not write standards in such a way that it will restrict or prevent the implementation of new and better solutions in the future.
“How you detect people within the zones around a mobile vehicle, today, it’s historically done with laser scanners and LIDAR,” he continues. “This whole revolution in 3D vision systems and better detection systems (with spatial awareness) will enable a lot of different things in the future, so we need to make sure the standards are able to accommodate those types of advancements and technology.”
“We don’t want to block a particular new technology that wasn’t known when the standard was initially developed,” says Gerstenberger. “We don’t want to impede technological progress. We want to enable it.”
With those technological advancements, will come more complexity. Especially when you have free-ranging robots and humans working in the same space.
NIST’s Marvel reminds us, “There’s no such thing as a safe or unsafe robot. It’s all about how you use it, how you integrate it. From there, you have to do your risk assessment.”
Risk Assessment or Not
That’s not a question. The latter is risky business, plain and simple. The R15.06 robot safety standard clearly states, “Risk Assessment shall be performed and is no longer optional.” It also states that the word “shall” is prescriptive, and describes mandatory requirements to comply with the standard.
“Risk assessment, risk assessment, risk assessment,” says Chris Soranno, Safety Standards and Competence Manager for SICK Inc. in Minneapolis, Minnesota, when asked what suppliers and end users should do in the meantime while the mobile robot standard is under development. “You can’t fix the problem unless you know what the problem is. With safety, the best way to identify the problem is to assess the risk.”
Soranno also serves on the R15.08 committee and has worked in robotics and the industrial safety market for his entire career, almost 20 years now. He’s seen a lot of applications.
“We work with our internal employees as well as our external customers to help increase competence in all aspects of safety,” says Soranno. “That includes where to find a standard, how to read and interpret a standard, and how to implement the requirements of those standards to achieve the objective of providing a safe workplace for employees.”
He describes the basic steps in the risk assessment process.
“You have to identify the tasks,” says Soranno. “You have to identify the hazards that those tasks are associated with. For each one of those, you need to identify the potential outcomes that would be a risk. And then rate those based on the worst credible level of severity of injury and the worst credible level of probability (how likely it is to occur). It could be based on frequency, exposure, number of exposures, awareness of people to the risks as they present themselves, or avoid-ability. Is it moving toward you at 100 mph or is it crawling really slowly so you can get out of the way?”
Who should do the risk assessment?
“It should always be a team,” says Soranno. “The stakeholders that have to live with the system, they will be the ones that know how they will interact with it, whether they’re operators, or maintenance personnel, or production people. They should be included on the front end. Then you have the equipment suppliers who are familiar with what the systems is capable of doing or not doing, and any other component suppliers, whether they’re safety component or end effector suppliers.”
In this mobile manipulator example, systems integrator and distributor Vicosystems marries a Mobile Industrial Robot (MiR) with a Universal Robots collaborative arm (pictured) to unload parts from shelving units and onto conveyors. Watch the video.
Changes Require Reevaluation of Risk
P&G’s Lewandowski says any new machine (or robot) that comes into the company requires a risk assessment.
“We always try to start with the integrators of the machinery or equipment, to have them do the first cut, or first risk assessment of any application. But once it comes into P&G, we have a process where we will then continue to build upon that risk assessment and even do a final risk assessment once that equipment or machinery is installed within P&G.”
If changes are made, then the risk assessment must be reevaluated.
“The expectation is that if you change the functionality of a machine or a process, that you review the risk assessment and make any changes or updates as required based on those changes,” says Lewandowski.
“Even if the environment changes, like the flooring, the rack placement, or the arrangement of the machines on the floor,” says Bollinger, then you need to reevaluate the risks. “What if you change one area of a warehouse into a manual picking station? That increases the risks, especially if you’re talking about a mobile robot in the space.”
The risk assessment must also address the environment in which the mobile robot is deployed. Soranno notes conditions such as the friction quality of the flooring, or how slippery it is, or the incline and smoothness of the floor.
He says it’s also important to have a process in place in case someone decides to change the environment, like repaint the floor for example, to make sure they don’t inadvertently change the properties that mobile robot is depending on to achieve its intended safety level.
“I’ve done thousands of risk assessments,” says Soranno. “There’s an exponential scale of how dynamic a system is versus how complex the risk assessment becomes. If you look at a power press, there’s typically only a handful of ways to interact with the machine. You identify those tasks, you identify the hazard associated with each of those tasks, and you come up with some remediation solution for how to reduce those risks to a level that’s acceptable or tolerable.”
Apply that process to a dynamic robot mounted on a mobile platform and it becomes much more complex.
“When you move to a robot, which is drastically more dynamic then a press and it can be used any number of ways, and not only that but it can be modified after it’s deployed to be used in new ways, that increases the complexity of the risk assessment,” says Soranno. “Now you take that dynamic robot and put it on a mobile platform that could be anywhere in a facility, and again it’s an exponential curve. You now have many more potential tasks or interactions.
“You’re not just worried about people walking up to the machine that’s bolted to the floor. Now the machine is moving around the facility and can drive up to a person that has their back turned to it,” he adds. “The complexity of that process gets a lot more difficult.”
Hail to RIA Volunteers
Those complexities will come under much deliberation by the R15.08 committee as they draft guidelines for the mobile robot space. You might be surprised to learn that the committee members are all volunteers, with day jobs.
The members meet together as a large group to hash out the details just three times a year, with smaller conference calls and meetings, and many drafts, in between. The next meeting is scheduled for November in conjunction with the Collaborative Robots and Advanced Vision Conference (CRAV) in San Jose, California.
Ask any of the members and they’ll tell you how they were inadvertently recruited for a spot on the standards committees.
Lewandowski jokes, “Usually you get involved by walking too close to a meeting or someone involved with a meeting.”
In other words, you get sucked in. But in all seriousness, they realize the importance of their involvement.
“By participating in the committee, we can be out front with the thought leaders,” says Bollinger. “It’s a benefit to us to understand where the technology is going and be brainstorming that with the other folks to understand the opportunities and challenges.”
“We think it’s very important and we have a passion for what we do,” says Lewandowski. “Not only for our company, but also to help the industry as a whole.”
Franklin echoes the significance of their invaluable contributions, “We greatly appreciate our committee volunteers’ time, energy and expertise.”
RIA Members featured in this article:
Bastian Solutions, LLC
Clearpath Robotics Inc. (OTTO Motors)
Fetch Robotics, Inc.
KUKA Robotics Corporation
Mobile Industrial Robots (MiR)
Procter & Gamble