How Standards are Contributing to the Future of Robotics
Humankind has long been fascinated by what we would today call robots. Stories are found across the world, from the singing automaton of the Taoist Liezi and the bronze giant Talos who patrolled the shores of Crete – to Sanskrit legends of machines that guarded the Buddha’s relics and the golems of Jewish folklore.
The Industrial Revolution transformed technology, and it reprogrammed our collective imagination. For the first time in history, mechanization offered the hope of finally ending toil and removing workers from hazardous roles. This situation was also fodder for science fiction writers, as manufactured thinking machines gradually replaced the automatons of ancient history.
While fantasies of dreaming androids and A.I. uprisings have yet to materialize, robots are increasingly part of the contemporary social landscape. Whether in manufacturing, packaging, transport, search and rescue, or space exploration, their pervasiveness suggests they will be part of humanity’s next great odyssey.
The prospect of a robot-powered future is exciting, but the technology is still in a period of growth. There are lingering questions about robots, from the tips of their fingers down to the bottoms of their feet. Today, standards are playing a pivotal role as roboticists look for solutions, especially by introducing rigor and repeatability into testing. The committee on robotics, automation, and autonomous systems (F45) is taking the lead, working to lay the groundwork for this emerging technology.
Locomotion is one of the basic concerns of robotics, and robot movement has been researched for many decades. For robots to be effective, they need to be able to get from point A to point B, and to do so in an intelligent way. Additionally, robots will increasingly need to be able to navigate environments that are shared with human beings – without becoming a hazard in the process.
These issues motivate a proposed standard, test methods for disturbance rejection testing of legged robots (WK86916). According to Adam Norton, associate director of the NERVE Center at the University of Massachusetts Lowell and chair of F45, legged robots have certain benefits compared to wheeled or track robots, but they also present unique challenges.
“There are a lot of advantages to legged systems that ground systems do not have, like walking up, over, or around obstacles,” Norton explains. “They introduce an additional level of complexity. If it can’t balance, then we can’t even talk about it doing anything else. This is a kind of enabling capability to allow other tasks to be performed.”
Dr. Bowen Weng, research scientist at Transportation Research Center on assignment to the National Highway Traffic Safety Administration (NHTSA), adds that there is significant work needed to establish a baseline for testing and verifying legged robots.
“The problem with legged robots is that we don’t have a well-established notion of safety from a standardization point of view,” Weng says. “When I got involved with ASTM, I started listing out work items we might want to figure out. What is even being tested? Are you testing them as a whole? Are you going to differentiate by categories? Because in vehicles, we have different categories, such as light vehicles, heavy trucks. What kind of equipment will be used to make it repeatable? How do you set parameters? Everything is up for discussion.”
Read More: A Road Map to ESG
In posing these questions, Weng drew on his experiences with vehicle safety, which showed what the automotive industry learned throughout its history. However, Weng appeared on ASTM’s radar after an Ohio State News article highlighted his study on improving testing of legged robots. Weng says that within 30 minutes of the article’s publication, he was contacted by Aaron Prather, director of ASTM’s Robotics and Autonomous Systems Program.
If approved, the standard that grew out of their conversation would be unique. “This is something that has not been done before,” Weng says. “Think about how the community has been doing tests. Just go online and watch YouTube videos. We have been rope-dragging, hand-pushing, foot-kicking, stick-poking, you name it. But none of those methodologies are standard ways of testing. They’re not repeatable, they’re not rigorous, and you cannot really have them as a benchmark performance characterization.”
Hold Out Your Hands
Weng’s comments show an important contrast between the high potential of robotics to transform the world and the elementary nature of many questions posed by researchers and developers. YouTube is full of both awe-inspiring success stories and “robot fails” compilations, demonstrating that issues affect every
Robotic hands, also called “end-effectors” or “manipulators,” are a key example. Whether on the assembly line or in potential everyday applications, a robot is not particularly effective if it can’t grasp objects and do so in an appropriate way. It was this matter that factored into the creation of the subcommittee on grasping and manipulation (F45.05), followed by a proposed test method for grasp-type robot end-effectors: grasp strength performance (WK83863).
Robot manipulation presents several levels of complexity. One level concerns the evaluation of end-effectors themselves. A second level addresses the more intricate actions of robot arms used in manufacturing, such as peg-in-hole insertions or nut threading. The highest level is mobile manipulation, introducing moving systems so robots can function in more advanced ways.
Norton says that it is crucial to start with the fundamentals. “If you’re going to grasp an object and do something with it, you need to know something about your end-effector before you can even think about what tasks the arm is capable of doing,” he says. “For some applications, this might be grasp strength in terms of maintaining a strong grasp, like if you are grabbing a very heavy object. Or your grasp needs to be solid but you’re grasping something very fragile, such as an egg. You still need some understanding of what this hand is capable of doing so that it can maintain a solid grasp without breaking or deforming the object.”
While simply looking at robotic end-effectors might sound like an easy task, getting it right has high stakes for the robotics community. There are many factors to consider, from the way a robot identifies an object and stably picks it up to how it moves in an environment with the object and manipulates it once it arrives at a destination. Selecting the wrong tools has the potential to be costly, and as the industry evolves, mistakes may generate hazards.
At this stage of development, Norton says there is a need for standards, but also a need to enable effective experimentation and innovation while the field is expanding.
“Standards play a really important role because you have this diverse landscape of technology, particularly in an emerging field,” he says. “We’re not interested in dictating the way that people develop their technology. We don’t want to turn off technologists, but we do want to make sure that we are able to compare them in an effective way or at least communicate in a way that is informative to the consumer, particularly around something like safety. This is filling a necessary gap, even more so because we’re still so early.”