Preparing for the evolution
The day when humans and robots work side by side is coming. Companies must prepare for this inevitable evolution in the workplace to ensure people will be safe in this environment.
Maybe 10 feet or so below an elevated platform, a homogeneous group of onlookers dressed to the nines—suits, ties, hats, gloves, dresses and blazers abound—seem mesmerized by the novelty that hovers above them. The way they look up at the behemoth, almost transfixed with their necks strained, one could easily misconstrue this curious group as people ready to meet their future overlord.
But after witnessing a 7-foot-tall, 265-pound, gold-colored metal “man” speaking hypnotically and moving methodically, who could blame them for gawping?
This scene took place at the 1939 World’s Fair in New York— less than two decades after Czech writer Karel Čapek first used the term “robot” in his play that has been translated as “Rossum’s Universal Robots.” The “World of Tomorrow” motto was showcased at the fair that year in a massive 1,200- acre area at Flushing Meadows-Corona Park in Queens. But the protractor- shaped platform inside this exhibit, no bigger than about 50 or 60 square feet, becomes the center of attention in a promotional video shot in “Wizard of Oz”-like Technicolor that is certainly not as vibrant as the film that also debuted that year and gave Judy Garland her signature role as Dorothy.
Not in Kansas Anymore
Elektro, the Westinghouse Moto-Man, is an imposing aluminum robotic figure. “He” walks slowly, tells corny jokes systematically and “smokes” a cigarette oddly when commanded by his human handler who communicates with “him” via a telephone handset that was, of course, technologically advanced for the time.
This robot, in essence, is the sci-fi fantasy come to life—a giant, intimidating, emotionless creation that really has limited use to humans but is certainly something to gawk at, much like a similar British Tin Man-like robot named Eric that debuted in the 1920s. In a separate video, Elektro professes, in a way that would make even The Robot from “Lost in Space” cringe, that “he” was designed to be subservient to humans “if you use me well.” This makes perfect sense in the context of Čapek’s coining of the term “robot” based on the Czech word “robota,” which means “forced labor.” However, it’s hard to believe Elektro’s words here, which, of course, were given to him by humans long before artificial intelligence (AI) was a Sirious—sorry, serious— specification for smart devices and other forms of technology. This robot was created during the dawn of the Asimovian era when the famous dystopian author Isaac Asimov wrote “The Three Laws of Robotics”.
Compare the scene from the Elektro video to a YouTube video of a 2018 ASIMO demonstration from the USA Science & Engineering Festival in Washington, D.C.
There, a diverse, casually dressed group of people, smartphones at the ready to record the moment, were introduced to the latest version of the Honda- manufactured robot.
In this demonstration, the human accompanying ASIMO towers over the diminutive 4-foot-3, 119-pound robot with the androgynous voice. The man also seems to command the expansive demonstration space while the robot acts as the eye candy. ASIMO then proceeds to show off its agility by climbing stairs, hopping on one foot, jogging and dancing by “constantly adjusting its center of gravity with every step it takes just like humans,” as the man explains. While Elektro unwittingly played the role of the Great and Powerful Oz to perfection, ASIMO comes across as more of a human companion, a friend, a helper, a Toto. He also seems to be about as intimidating as the Cowardly Lion at the deli counter sheepishly dreading the moment when his number is about to be called—a good quality for a collaborative ’bot.
“The futuristic world of robotics is a fascinating case study for workplace safety,” said Paul Marushka, Sphera’s president and CEO. “While robots are designed to go where workers shouldn’t and handle things that could otherwise put a person in harm’s way, companies must also prepare for the evolution of the technology when robots and workers do their jobs side by side. We’re on the cusp of a very exciting time, but true industrial progress in robotics means safety must come first. An Integrated Risk Management approach is vital to Operational Excellence for human- robot collaboration.”
Today, the “World of Tomorrow” is on full display in the industrial world with advancements in robotics, and will continue to evolve over many future tomorrows. The truth is the robots aren’t coming; they’re already here and have been for decades. Distance is the difference-maker in the industrial setting. Since they entered the workforce, robots have been mostly separated from the humans that work around them by some form of barrier; it’s just safer that way. In the future, industrial robots will be uncaged and ready to work side-by- side with their fleshy fellow workers.
The truth is the robots aren’t coming; they’re already here and have been for decades.
General Motors is credited with being the first automotive company to incorporate the Unimate robot into its assembly line for diecasting in the early ’60s, but outside of perhaps some of the high-end vehicles in the marketplace today, automakers now universally employ robots for precision and speed in the mass production of automobiles, and especially to keep workers out of harm’s way.
Beyond automobiles, we are seeing industrial robots in many different settings—from the “iron roughnecks” in the Oil & Gas industry that connect and disconnect drill pipes to the cartesian robots in the chemicals industry that move dangerous substances with precision so humans aren’t at risk of exposure from spills to British Airways’ new mototok robot that can tow a 130-ton plane with ease while reportedly reducing pushback delays by 54 percent. Heck, there’s even a company in California called Creator that recently launched a burger-making robot that grinds and cooks the meat to order and slices and dices and drizzles on the fixin’s to assemble what Bloomberg calls “the freshest burger on Earth” in five minutes for six bucks.
Also, Italy’s IIT-Istituto Italiano di Tecnologia is currently working on a firefighting robot. It’s remote- controlled Walk-Man ’bot—and, yes, it’s probably time we quit recycling names with these robots—is designed to walk into a building and use a fire extinguisher to put out a fire.
Another example worth noting is how GM returned to its robotics roots a little over a decade ago when it teamed with NASA to create Robonaut 2, a humanoid ’bot, which had “lived” on the International Space Station since 2011. As NASA explains in a fact sheet about the robot, “The conditions aboard the space station provide an ideal proving ground for robots to work shoulder to shoulder with people in microgravity.” Eventually the goal is to get Robonaut 2 or its future progeny out and about to work in the vacuum of space by servicing satellites and to boldly go where no one has gone before by exploring asteroids, comets and perhaps even roving on Mars one day. Unfortunately, NASA recently had to bring the robot back to Earth for repairs following a few years of unsuccessful attempts at getting Robonaut 2 to work with its legs, which were added in 2014. Alas, sometimes progress can indeed be lost in space.
With the global market for industrial and nonindustrial robots expected to jump from $39.3 billion in 2017 to an out-of-this-world $500 billion in 2025, according to Statista, the evolution of the workplace will not be televised as much as it will be roboticized.
As part of that evolution, many have anthropomorphic qualities to them, like “hands” that grip and twist, but they don’t often look like humanoids. With innovations in robotics and advancements in AI and machine learning, that’s likely to change. One of the goals of robotics going back even before Walt Disney’s Great Moments With Mr. Lincoln—a forerunner to Disney’s current Hall of Presidents exhibit—debuted at the 1964 World’s Fair was to create androids that act and look like humans but offer precision, data processing power and an indefatigable ability that, for lack of a better phrase, is not at all human.
Robots like Honda’s ASIMO and Kawada Industries’ HRP-4 robots couldn’t be confused for the humanlike Replicants from “Blade Runner,” but companies like Hanson Robotics and Hiroshi Ishiguro Laboratories are moving in that direction with more human-looking robots like Sophia and Erica, respectively.
As technology improves, we are continuing to move closer to a time when it will be commonplace for robots and humans to work together and even collaborate.
So what are the potential safety implications of this new-look workplace where man and machine are co-workers, and what can companies do to ensure the safety of their human colleagues when the future becomes the now?
As the U.S. Occupational Health and Safety Administration (OSHA) notes on its website, “Studies indicate that many robot accidents occur during nonroutine operating conditions, such as programming, maintenance, testing, setup or adjustment. During many of these operations, the worker may temporarily be within the robot’s working envelope where unintended operations could result in injuries.”
It’s not that surprising that the case studies of robot-related accidents on OSHA’s website are few and far between—and mostly quite old. Indeed, one of the reports from the National Institute for Occupational Safety and Health describes an incident from 1984 where a worker went into cardiac arrest after being pinned between an industrial robot and a steel pole in a document titled “Request for Assistance in Preventing Injury of Workers by Robots.” Another case study from 1999 discusses a young worker accidentally tripping a light sensor in a meat-packing plant causing a robotic platform to crush him.
Overall, OSHA’s website lists only 38 robot-related accidents between 1984 and 2016, but 27 of those accidents (71 percent) were fatal. Incidentally, the first documented robot-related fatality took place in 1979, five years before this reporting period, when Robert Williams was killed by a parts retrieval robot. It is worth noting that a jury awarded Williams’ family $10 million a few years later (about $25 million in today’s dollars) from the robotics manufacturer because the jury found that safety measures, including an alarm, were lacking.
More recently, an Arizona woman crossing the road with her bicycle was killed when an Uber self-driving car ran into her. (Self-driving cars are considered robots because, by definition, they are “mechanical device[s] operated automatically” even though humans are still supposed to monitor the roads as a failsafe.) The accident is under investigation and Uber has reportedly settled with the woman’s family privately, but this is the first fatal accident involving a self-driving vehicle.
Another woman, who was 20 years old at the time of her accident, was killed in 2016 when she entered a robotics station to clean a sensor and was crushed when the robot unexpectedly restarted. The company was fined $2.5 million by OSHA, and a separate lawsuit from the family is pending.
These incidents, aside from the self-driving vehicle collision example, took place in situations where humans and robots were not supposed to be working closely together. Without sounding alarmist, this is why it is so important for companies to be prepared when people and ’bots work side by side on a greater scale.
Together at Last
In 2016, the International Organization for Standardization (ISO) released its ISO/TS 15066 standards for human-robot collaboration. At the time, Carole Franklin, the secretary of ISO/TC 299/WG 3, said in a written statement that the new standard “gives specific, data-driven safety guidance needed to evaluate and control risks.”
The article announcing the launch of the new standards also explains the difference between an industrial and collaborative robot this way: “Collaborative robotics is when automatically operated robot systems share the same workspace with humans. It therefore refers to a system or application rather than a particular type or brand of robot. Industrial robots do not include an ‘end effector,’ that is, the device at the end of a robotic arm that interacts” with its environment.
Somewhat eerily, according to the article, “The technical specification includes data from a study on pain thresholds of different parts of the human body. This information can be used to develop and implement collaborative power- and force-limited robot applications.”
The power, in other words, is in innovation.
In 2017, a group of robotics researchers published a paper in the International Journal of Advanced Robotic Systems called “Human-Robot Collaboration in Industrial Applications: Safety, Interaction and Trust” that details takeaways from the European Union FourByThree Project, which ended in 2017. As the report explains, “Safety is one of the most critical aspects in this collaborative human-robot paradigm.”
To achieve a safe, cohabitated working environment, “the robot must be constantly aware of what is happening around it, and it has to monitor the workers’ actions in order to change its behaviour (speed and/or trajectory) according to the separation distance.”
The four main characteristics of the project included: modularity, i.e., hardware and software tools for developing robotic solutions; ease of use to facilitate robot programming; efficiency, i.e., making robots reliable and maintainable; and safety.
“The safety approach is centered on the design of the actuators,” the article says, “with the capability to monitor the force and torque in each one, which provides the opportunity to implement variable stiffness strategies and reactive behaviour in case of contact/ collision.”
As the website RobotShop explains, a robotic actuator is any device that converts energy into physical motion, such as a DC motor.
In an email, Iñaki Maurtua, the autonomous and smart systems manager at IK4-Tekniker, a research center for advanced manufacturing and information and communications technologies based in Spain, who was the lead researcher on the “Human-Robot Collaboration” report, said: “FourByThree has developed a set of components that help create a new generation of modular industrial robotic solutions that are suitable for efficient task execution in collaboration with humans in a safe way and are easy to use and program by the factory worker.”
The hardware and software elements, he said, provide for custom robotic solutions including robotic arms and safety mechanisms that allow system integrators to let robots interact with people effectively
and safely. He added that the thing that surprised him most about the project was how open people were to working closely with robots once all the safety measures in place were explained to them. As with any complex piece of technology, training is key to ensure robots and people can work hand in hand.
Additionally, in the future, “AI will be an interesting tool to facilitate (reduce) the need of programing robots and create robots with human-aware behaviours,” Maurtua said.
It will also require a path to understanding each other. When a worker tells a robot to do something, it’s critical that a robot can “comprehend” what is being asked of it to ensure not only that the task is completed to the worker’s satisfaction but also to avoid unintended incidents from a failure to communicate between human and robot and perhaps between robot and robot as well.
Communication works both ways by the way. That’s why Rethink Robotics, for example, includes cartoonish moving eyes and eyebrows on its Baxter and Sawyer robots that allow people to see where the robot is about to maneuver when they are performing a task such as machine tending, loading and unloading, molding, packaging or even playing a game of tic-tac-toe as part of the Museum of Science & Industry Chicago’s Robot Revolution exhibit.
“The FourByThree project has designed and developed a semantic approach that supports multimodal (voice and gesture based) interaction between humans and robots in real industrial settings,” according to the “Human-Robot Collaboration” report, which includes a voice interpreter, a gesture interpreter, a fusion engine—a combination of the two that could be important in a noisy setting—and a knowledge manager that “describes and manages the environment and the actions that are feasible for robots in a given environment, using semantic representation technologies.”
But understanding commands is one thing and ensuring robots don’t unintentionally hurt humans is another.
One way that safety goal might be accomplished is through motion capture. Companies like PhaseSpace create motion capture through active LED technology that could help humans and robots work together in close quarters. In a video from the Massachusetts Institute of Technology titled “Analyzing the Effects of Human-Aware Motion Planning on Close-Proximity Human-Robot Collaboration,” a researcher wears an LED glove and interacts with a robot by screwing in boltlike objects and then having a robot tap the bolt only after the human hand has moved.
Another MIT report called “Efficient Model Learning from Joint-Action Demonstrations for Human-Robot Collaborative Tasks” from a few years ago, details how a robot held a box and interacted with a researcher wearing a similar glove.“ Measurements of the hand position were averaged in fixed intervals, and an action was detected when the difference between two consecutive averages exceeded a set threshold.” Eight cameras were used to track the motion of the glove.
Separately, German robotics manufacturer KUKA says on its website that its LBR iiwa robot is the “world’s first series-produced sensitive, and therefore HRC-compatible, robot.” A video demonstration on KUKA’s website shows a man having a conversation with a woman and “accidentally” putting his hand on the assembly. When the robot touches the human hand, it automatically comes to a stop and recoils from the extremity.
Speaking of limbs, a Japanese company called Cyberdyne has developed a cyborglike robot it calls HAL, which shares no resemblance to the nefarious computer from “2001: A Space Odyssey.” This ISO 13485 (medical device)-certified robot is designed to help people walk. According to the Cyberdyne website, “HAL for Medical Use— Lower Limb Model moves the wearer’s legs in accordance with the wearer’s intention, ‘I want to walk’ or ‘I want to stand up.’ ” While this is listed as a medical device, it’s easy to see the possibilities for industrial usage as well in assisting workers get around who otherwise would not be able to or would not be able to easily because of a medical condition. Similarly, as ZDNet recently reported, Ford Motor Co. ordered 75 exoskeleton vests from another manufacturer to give to some of its factory workers to help reduce the strain of lifting.
If I Only Had a Brain
The Scarecrow from “The Wizard of Oz” sought the Wizard’s help in finding him a brain. In robotics, that is the equivalent of the search for AI and machine learning. It’s an interesting field for sure, and one that is constantly changing.
A recent MIT Technology Review article acknowledged as much in a piece that explores how “AI and robotics have been separate fields up to now. Combining them could transform manufacturing and warehousing—and take AI to the next level.”
Believe it or not, the example the article leads with is a robot picking up karaage chicken (the Japanese take on fried chicken).
It might sound simplistic, but as the article explains, the Osaro FoodPick does something that most other robots can’t. It uses machine learning and AI to figure- out how much force is necessary to grip and sort the food without damaging it. And quickly, too. After all, a chicken chunk would go splat pretty quickly if gripped too tightly as would a human appendage that comes in contact with a robot with an overzealous gripper. According to the article, “An ordinary robot arm lacks the sense needed to pick up an object if it is moved an inch. It is completely hopeless at gripping something unfamiliar; it doesn’t know the difference between a marshmallow and a cube of lead. Picking up irregularly shaped pieces of chicken from a haphazard pile is an act of genius.”
These advancements are part of an evolution in AI and machine learning that has yielded a subset area of research called “deep learning” where a robot learns on its own without human intervention.
That deep learning was on full display in 2016 when AlphaGo, an AI program, defeated Lee Sedol, a champion of the Chinese board game Go, four games to one. As Amit Gupta, the chief information officer at the American Institute for Chemical Engineers, wrote in the June 2018 issue of Chemical Engineering Progress: “Experts believed that computers would not be able to defeat top human players at Go for decades, if ever During the games, AlphaGo played highly inventively, making moves that no human ever would, and defeated the 18-time world champion.”
The article also explained how Shell uses AI to predict valve failures. “Predictive analytics and machine learning algorithms could make it possible to avoid unexpected failures and unnecessary maintenance,” Gupta wrote, “which would save millions of dollars per year in optimized maintenance and deferment avoidance.”
Predictive analytics is a recent development in workplace safety. It’s a way to focus on the leading rather than lagging indicators to help prevent incidents before they take place. To get there, a company would need data, lots of it, as well as powerful software tools to be able to harness it. While such analytics wouldn’t be able to cite when exactly an incident is going to take place, they would be able to point out areas of a workforce that are most at risk so that issues can be addressed before something bad happens.
As Hariharan Ananthanarayanan, a robotics engineer from AI software company Osaro, wrote on the Re-work website: “Technologies such as machine learning and deep learning are providing robots with the autonomy required to recognize their environments and define their tasks.”
With these types of advancements in AI, machine learning and deep learning, once robots are thrown into that mix, you have the perfect complement of predictive analytics and muscle to allow companies to benefit from the technology at their disposal while keeping workers safe. The “smarter” robots are, the more helpful they become, especially when they are needed for use in areas that are considered unsafe for humans.
“For example, robots are already used to perform repetitive and monotonous tasks,” according to a discussion paper from the European Agency for Safety and Health at Work, “to handle radioactive material or to work in explosive atmospheres. In the future, many other highly repetitive, risky or unpleasant tasks will be performed by robots in a variety of sectors like agriculture, construction, transport, health care, firefighting or cleaning services.”
Companies have to make sure the safety of human beings is never compromised.
Just imagine how valuable a robot like the aforementioned Walk- Man would be if it could find and confront danger on its own.
Of course, Narenderpal Marwaha, Sphera’s principal solution architect, warns that with AI comes the need for added cybersecurity to ensure a robotic worker isn’t hacked.
“The key differentiator of the security risks is that the damage is intentional. For process industries, SVA is used to identify security threats—as opposed to the inadvertent process and occupational hazards—assess attractiveness of the target to an adversary, and evaluate the vulnerability of the asset to an attack,” he said, referring to Security Vulnerability Analysis. “The term ‘Cryptobotics’ has been proposed to study security measures in robotics, and companies will need to take this potential risk very seriously.”
In a 2015 research paper titled “Cryptobotics: Why Robots Need Cyber Safety,” the authors explain, “Among the common attacks computers may suffer, there are: denial-of-service, eavesdropping, spoofing, tampering, privilege escalation, or information disclosure for instance. To these problems, robots add the additional factor of physical interaction. While taking the control of a desktop computer or a server may result in loss of information (with its associated costs), taking the control of a robot may endanger whatever or whoever is near.”
As the article explains, many ’bots use the popular Robot Operating System architecture, but it cautions that the nodes used for communication do not know whom they are communicating with. To deal with that problem, developers are incorporating Data Distribution
A time when humans and robots work together more closely is inevitable. How soon that will be is debatable.
Service (DDS) into ROS 2.0 as a “transport layer.” As the authors explain, “It would be a big step forward for securing our robots if ROS 2.0 aimed to comply with the DDS security specification as well,” which includes authentication, access control and cryptography.
In Spite of the Wonderful Things They Do
Without a doubt, robots are tools that can make work easier and safer for humans, but that does not mean they don’t come with their own risks. Besides the cybersecurity and other safety issues mentioned in this article, there are other risks worth noting.
“A robot could suddenly lose its power or not know how to deal with unprogrammed situations leading to disaster,” said Subha Krishnamurthy, Sphera’s senior product manager. “Companies have to make sure the safety of human beings is never compromised.”
Additionally, “Robots lack one very important skill: judgment,” said John Crosman, Sphera’s senior process safety consultant. “Robots may not necessarily be able to assess information from a new, previously unidentified situation to determine whether a hazard could be present, and may therefore not be equipped to react appropriately. This means that the importance of conducting and regularly updating the various safety studies and risk assessments will take on additional significance since the machines will only be as good as the knowledge programmed into them.”
Another counterintuitive risk is an overcautious robot, Crosman said. “I am worried about robots being overly analytical and overly cautious, initiating a potentially higher number of production interruptions than we see today. While I am never against tripping a process in the event there is the potential for a hazardous event, tripping too early or too often inadvertently puts the plant into one of its most hazardous states— starting it back up again.”
Until we figure out how conservative robotic developers need to be to balance safety with production demands, Crosman said, “there is a potential for a lot of spurious or unnecessary production interruptions that could frustrate the integration of robots into the production process.”
Maintenance is also a big issue, said Francis Trudeau, Sphera’s solution manager for Product Stewardship. Like a shiny new car that gets driven home from the dealership, collaborative robots fresh off the assembly line can start out as state-of-the-art “Rolls Royces” but become the “Yugos” of the production floor if not properly maintained. “When AI-powered- robots hit the workplace, a whole domain of responsibility opens up where you will need to have an industrial engineer of sorts who is responsible for caring for the machines and ensuring human safety first,” he said.
Just Doing My Job
While the fear of robots taking over the world has been written about for decades now—even “The X-Files” addressed this in the standalone episode titled “Rm9sbG93ZXJz” (translated as “Followers” in Base64 code) when devious sushi robots declare war on Agents Mulder and Scully over a tip or lack thereof—the bigger concern many people have is losing their jobs to robots.
Many experts believe the influx of robotics technology will lead to a change in the marketplace much as computers did when they entered the workforce en masse in the 1980s. A recent McKinsey & Co. report estimates 15.8 million jobs were created in the United States since 1980 from the introduction of the computer—“even after accounting for jobs displaced.”
Perhaps more eye-opening is that McKinsey predicts that 20 million to 50 million new jobs could be created in tech fields by 2030.
“People work best when they are challenged, engaged and present,” said Philippe Guillard, Sphera’s director of solution consulting. “It’s no secret that, when people get disinterested, their work ethic and quality fade fast. While this is already happening to a degree, use your people to figure out how to solve challenges, and then enable a robot to repeat the solution for actual production. It’s not a question of if a person will make an error in a repetitive task but when. People were just not designed to do repetitive tasks.”
And as Dominik Boesl, vice president of consumer-driven robotics at KUKA, told Popular Science, “We don’t want to replace people; it’s more about enhancing and augmenting people.”
A time when humans and robots work together more closely is inevitable. How soon that will be is debatable; some say sooner and some say later. Regardless, it’s incumbent upon companies to be prepared for when that day comes. Even Elektro wouldn’t argue with that.