The human eyeball as a primary means for tracking inventory on retail store shelves may soon go the way of the dinosaur.
That is a bet Richard Schwartz of Pensa Systems Inc. is counting on for the benefit of retailers, brands and consumers.
The founder, president and CEO of Pensa is pushing a system that combines autonomous drones, computer vision and artificial intelligence for in-store inventory management — employing emerging technology to do what humans have traditionally done.
Being out of stock on store shelves has long been a bane of the retail industry, with more than one in eight products often unavailable on shelves at any given time, Schwartz says. Retail analyst IHL Group estimated in 2018 that it was responsible for retailers losing $984 billion in revenue worldwide, including $145 billion in North America.
With shoppers often postponing purchases when the products they want are not on the shelves, “it has been a hard problem to deal with despite the huge revenue implications as well as the consumer frustration,” Schwartz says. “That’s a very material loss for the retailer on the revenue level and profitability level.”
At the core of the Pensa Systems platform is the front-end deployment in retail stores of small, camera-equipped drones that capture stock conditions on store shelves. The drones are self-guided and automatic, acting as “a roving pair of eyes for staring at the shelves,” Schwartz says.
The thought of drones guiding store-shelf inventory tracking seems cutting-edge enough, but Schwartz says it is on the back end where retailers can enjoy the primary benefits. Stock inputs captured by the drones’ cameras are forwarded in real-time and stored in the cloud, affording stores immediate visibility into their stock positions. The AI architecture then learns and analyzes information from the shelves, such as recognizing how products are managed while drawing conclusions about inventory status.
The drone operations are designed to be unobtrusive and quiet in retail store environments and, Schwartz says, are far less intrusive than recent generations of robots that have launched in retail stores for tracking inventory. Because Pensa Systems’ drone operations are automatic, the devices are set to deploy during hours when stores are closed or during store hours in areas where aisles are free of shoppers.
In testing the technology during store hours, Schwartz says Pensa Systems found nearly half of shoppers were unaware of the technology’s presence; those that did notice were more curious than put off.
He says such automation is overdue inside stores to join progress that has been made in automating retail warehouse operations and supply chains.
“Where it actually matters for a commerce transaction, there’s no automation at all. It’s been very manual and expensive. It is very people-intensive, with people that are literally going around and doing a visual audit and trying to stare at the shelf to see what’s out, to see what’s running low and to see what’s misplaced,” Schwartz says.
With AI, the system’s computers can be trained on the back end to recognize products on the shelves almost to the individual item level, he says; store shelves are designed inherently for products to be recognized easily by shoppers, and AI leverages that advantage.
“We train the computer not only literally to look at the shelves, but effectively to recognize things the same way people do,” he says. “It sort of learns as it goes.”
Pensa Systems is working with major retail chains and large brand manufacturers to move the technology into the store while focused on particular product categories, Schwartz says, such as consumer packaged goods.
M.V. Greene is an independent writer and editor based in Owings Mills, Md., who covers business, technology and retail.