Is there a place for human beings in the warehouse of the future?

Or, for that matter, in the warehouse of today? Around the world, there’s already a smattering of distribution facilities that require no human labor at all — machines are doing all of the putaway, picking, packaging and preparation of goods for shipment. (At which point, thankfully, there’s still a person waiting at the loading dock to haul them away — at least for now.) No need for breaks, lunch, climate control, sick days or PTO. Nor for illumination in the aisles — hence, the term “lights-out” warehouse.

Is this our future? Fueled by robotics and artificial intelligence, and following the trend in many factories, warehouses seem to be on a relentless march toward total automation. That’s not the conventional wisdom, however: Industry experts would have us believe that this people-free scenario will continue to be limited to a select number of sites and product types for the foreseeable future. They paint a picture of humans and robots working efficiently side by side, even if the traditional mundane tasks of physical handling become the sole responsibility of the latter. Still, the leaps and bounds that AI has experienced over just the last couple of years seem to be boosting the prospects of the dark warehouse.

There’s one place where lights-out warehouse operations are already making a splash: the night shift. In a number of facilities, automated pickers are eliminating the need for overnight staff, says Jan Zizka, co-founder and chief executive officer of Brightpick. His company makes robots that can pick and buffer orders without human intervention. Orders picked overnight are readied for packing and shipping at the start of the next regular human-staffed shift.

The robots are controlled by AI and feature mobile arms, equipped with 3D vision and force-sensing grippers, that can pick individual items from shelves and totes. Brightpick claims that the underlying software allows each robot “to see, think and act with human-like responsiveness and adaptability.”

Zizka says the technology lends itself especially well to pharmaceuticals, small electronics and grocery items. Accuracy can be as high as 99%, he says, as long as the labeling is clear and the system is dealing with familiar types of packaging. If it isn’t, then the robot can be trained to recognize and handle the novel configuration.

Today’s picking robots boost operational efficiency by selecting the shortest possible route through the facility. Figuring that out in a site with thousands of different products and orders can be brutally difficult — a variation on the old “traveling salesman” routing problem, contained within four walls. Zizka says the system can cut travel time in the warehouse by as much as 50%.

The deployment of robots in the warehouse isn’t limited to night work, of course. Automation has had a profound effect on everyday operations over the last few years. The “lights-out” horizon seems to be drawing ever closer, however, and AI is the game-changer. The average distribution facility consists of a huge variety of product sizes, package weights and floor layouts.

“It looks simple to a human who can handle any tote,” Zizka says, “but for a robot it's some kind of challenge.” A system that can automatically handle those complexities “wouldn’t have been possible 10 years ago.”

Zizka isn’t prepared to write the obituary for human-staffed warehouses just yet. For the most part, he says, people still need to show up in the morning to work the buffer that was set up by the automated pickers overnight. But the number of humans needed to do even that job is shrinking fast. For the future, Zizka envisions the need for a single “robot manager” onsite. He cites the old joke, credited to business consultant Warren Bennis, about the factory of the future, which will have just two employees, a man and a dog. “The man will be there to feed the dog. The dog will be there to keep the man from touching the equipment.”

Are warehouses on that same path? Zizka predicts that some major distribution centers will be fully automated within five to 10 years. “I don’t see any technical difficulty,” he says.

So let the dogs in.