Stop me if you’ve heard this one before: Labor is tight, but consumer demand is booming, so warehouse and DC leaders are turning to automation to keep up with the workload. That basic scenario has been playing out for years, with automated equipment vendors providing ever-more-powerful tools to boost fulfillment rates.
The latest round of warehouse tech includes robotic picking arms, autonomous mobile robots (AMRs), and artificial intelligence (AI). Add it all up, and the resulting combination can seem like a nearly human collection of hands, legs, and brains. Some forecasts even suggest that machines will soon replace people in the distribution center, creating a “dark warehouse” that needs nothing more than a reliable power supply to operate 24/7.
But even as robots become a familiar sight in warehouses across the country, experts say human workers continue to play an important role. That’s largely a reflection of the varied nature of warehouse work. Unlike an automated assembly line, where robots perform structured and repetitive tasks, warehouses demand a significant amount of flexibility—think of today’s e-commerce fulfillment centers, where items, quantities, and packaging vary from order to order, and demands change from shift to shift.
Given those complexities, a successful robotic implementation still requires the participation of humans—whether they’re working collaboratively with the bots to pick orders, handling errors, or supervising fleets.
When it comes to applications that integrate people and robots, most people think of the collaborative robot, or cobot—a robot that works alongside human workers in a semiautomated process that leverages the strengths of both. An example might be an AMR that can navigate its way to an assigned warehouse rack but lacks the ability to pick individual goods efficiently—a job that is then performed by its human “collaborator,” who selects the items and deposits them into totes on the AMR.
But in many cases, the human worker’s contribution to the operation is less physical than cognitive.
Compared with robots, people are more flexible in their thinking and better at solving complex problems, says Stephen Dryer, senior global product manager for the material handling systems integrator Fortna (which recently merged with MHS Global).
“All the things that robots are not very good at will be the purview of the human,” Dryer says. “There’s a fear that robots are going to take over people’s jobs, and it is absolutely the case that robots can do certain things pretty well. But they are not efficient at higher-order tasks”—particularly ones requiring judgment calls and problem solving.
Dryer compares the current state of warehouse robotics with what’s happening in autonomous trucking. “It’s like the self-driving story; there were predictions of self-driving vehicles taking over and of people not getting jobs in trucking—or deciding not to go into the sector. But we’re not seeing that. Because with driving, you still need human brains, human eyes, human decision making,” he says.
The need for that higher-order work has driven the development of “human in the loop” (HITL) robotic systems, also known as “brains in the background” systems. As opposed to working shoulder to shoulder with a cobot to pick e-commerce orders, a person working in an HITL system serves as a supervisor. HITL systems need people for the same reason that a computer printer that can produce hundreds of copies of precisely printed pages still needs a human to clear paper jams or replace an empty ink cartridge.
In the warehouse, an employee working with HITL robots will monitor operations on the DC floor, and when a problem occurs, quickly step in to resolve the issue and avoid a systemwide work stoppage, Dryer says. For example, that worker might notice an operational logjam or a dropped package—known as an “exception event”—and get the robot back on track by resetting it to its “home” position or returning the fallen box to a picking zone, he says.
Many DCs have dubbed these robot supervisors “water spiders,” a nickname derived from their habit of darting around the building the way a water spider scurries around a pond, fixing problems for robots, says Erik Nieves, CEO and founder of the parcel-handling robotics platform Plus One Robotics.
“What people are good at is decision making, dealing with exceptions as they happen, and using our cognition and flexibility,” Nieves adds. “And the warehouse is predicated on variability, not predictability. So the lesson is, ‘Thou shalt have a human in the loop.’”
The HITL concept originally grew out of cases where manufacturing facilities would assign people to repetitive tasks that were just slightly too complex for machines, termed “almost automatable,” Nieves says. “When a robot [encounters] something it doesn’t understand, a remote supervisor can step in and give it a command or show it what to do. If you can’t find a way to deal with exceptions, you are DOA, so you need to have HITL,” he says.
In Plus One’s case, that human in the loop is a “crew chief,” the company’s term for the remote supervisors who troubleshoot problems with clients’ automated systems. Available 24/7, these crew chiefs work in shifts from the company’s San Antonio headquarters, watching video feeds of warehouse robots in distant cities and putting things right—say, reorienting a confused robot—with the click of a mouse. Nieves notes that the job requires quick reactions and good judgment, making it suitable for someone with a background in computing or video gaming, but that it doesn’t require a college engineering degree.
That remote oversight allows the company to solve the majority of problems for warehouse robots, barring the rare physical problem, he says. “Occasionally a crew chief might see that a vacuum cup blew out, or a box broke open and there are DVDs all over the floor or something. Then they would alert a local person, usually staff from the maintenance department, and say ‘Cleanup in aisle 6,’” Nieves says.
Human workers play a similar role at Phantom Auto, a San Francisco-based provider of remote operation systems for forklifts. Drivers operate the vehicles from an office cubicle by viewing a live video stream from each remote-controlled lift truck, via a system that provides a 360-degree view and two-way audio. Like Plus One’s crew chiefs, they resolve the occasional physical problem inside the warehouse, known as an “edge case,” by notifying an employee in the building, says Elliot Katz, Phantom Auto’s co-founder and chief business officer.
Katz scoffs at the idea that warehouse robots will someday replace human workers entirely. “When the pandemic hit, people couldn’t go to work in close confines. And if AMRs were fully functional, that wouldn’t have been a problem. But that automation still can’t handle complex environments. There was never a better time for ‘robots to take over our jobs’ than the pandemic, when we couldn’t even go in the building. And it didn’t happen.”
As robotic technology continues to improve, autonomous platforms will take on increasingly complex tasks. But their abilities will always fall short of humans’ capacity to solve problems with creativity, Katz says. “‘Fully autonomous’ doesn’t exist. Robots are always going to be cobots,” he explains. “If and when robots start taking on [expanded roles] in larger deployments, that will just create new jobs for people as you have to have humans overseeing the operation and intervening when there’s an edge case.”