GNC’s AI drone catches employee oversight and reduces inventory errors

AI For Business


At the GNC warehouse in Whitestown, Indiana, employees can be seen and heard overhead every day, counting and tracking entire boxes of product across more than 2,000 pallets in the warehouse aisles, a task that few humans are willing to perform.

Bill Monk, GNC’s vice president of distribution, said regular drone flights allow nutritional food manufacturer GNC to get a complete picture of its 250,000 square feet of spare inventory once a month instead of once a quarter. More frequent monitoring will help the company achieve its goal of reducing backorders. A backorder occurs when an item is not on the shelf or completely missing from the warehouse, but the inventory system records it as being in stock.

“This allows for a faster cycle. If you don’t know the location, you can’t ship it,” Monk told Business Insider.

GNC started using drones made by Corvus Robotics two years ago. Since then, the number of unshipped items has plummeted from a few hundred items per day to about 98 items per day, Monk said. GNC inventory personnel now have more time to investigate and track discrepancies by validating and remediating issues flagged by the drone system.

AI helps with drone positioning and insights

Jackie Wu, founder and CEO of Corvus Robotics, said GNC’s inventory staff works with Corvus to coordinate drone schedules so that the drones can cover specific areas during 25- to 30-minute flights.

“There’s a lot of physical AI in the vehicle that will allow it to collect data and move autonomously in these environments,” Wu told Business Insider. “AI helps customers with what to do with inventory data and how to handle discrepancies.” Specifically, AI helps drones obtain accurate locations so they can move safely and reliably. Drone images and videos can also help staff find misplaced products, half-loose pallets and overturned boxes.

In general, drones can connect to networks and access the internet to remotely query large-scale AI models, said Brendan Englott, director of the Stevens Institute for Artificial Intelligence and a professor at the Stevens Institute of Technology. He said the networking capabilities allow the drones to operate with less onboard computing, allowing them to fly longer before needing to be recharged. Warehouses also have stable lighting conditions and few or no windows, allowing the drone’s computer vision algorithms to be tailored to a single environment, which is ideal for performance, Englott said.

AI also makes it easier for non-experts to ask questions and provide drone instructions in plain English, Englot said. “The language elements of these models bridge the gap between professional robotics engineers and warehouse employees,” he said.

Drones reduce operational challenges for GNC

Before using drones, Monk said, GNC had an inventory staff of 20 people, but it was difficult to find people to do the grunt work and pick up the undesirable shifts required for manual inventory counts.

Monk said employee turnover has improved since the company introduced drones. Monk said he was happy to do inventory by drone because “it’s not a very fun job and you have to count a lot of pallets to find the needle in the haystack.”

Tammy Lutcher, senior inventory management specialist at GNC, agrees. “The drone handles the tally and allows our team to do more investigative work. This takes a lot of the effort out of the job,” Lasher said. “It’s really cool to work with this kind of technology every day.”

With the introduction of drones, most of GNC’s original inventory team members have been reassigned to various roles, such as customer service and inventory accuracy in areas with high product picking, or are attrition-driven, Monk said. Currently, one or two inventory team members are focused on the reserve inventory area and regularly verify the accuracy of drone data. “We will definitely audit the 2% of what the drone says is correct,” Monk said.

Once, a drone showed a location with 600 boxes, but this contradicted GNC data, which recorded 60 boxes. “Everyone thought drones were wrong,” Monk said, but that wasn’t the case. A staff member entered the wrong box size and the trailing zero did not appear in the measurement.

Drones have some limitations

Drones have some limitations. For GNC, one of those is plastic-wrapped pallets, Monk said. GNC’s reserve inventory is located in a 70-inch aisle, a tight space for a drone whose rotors are two-thirds that wide. The drone’s sensors can see through the plastic to count inventory, but if the plastic breaks and flaps, the blade could get caught and the drone could fall, he said.

To avoid this, our staff are trained to cut and remove plastic when removing boxes from pallets. Monk said workers typically inspect the corridor for problems before a scheduled drone flight.

He added that while GNC’s accuracy has improved significantly, the company is only using its drones for back-up inventory (which typically includes unopened boxes). Drones can’t look back at an opened box to see what’s inside, but the software can provide case-level estimates, Wu said.

Still, the drones helped GNC in ways Monk didn’t expect. “If you don’t find a problem after a quarter or six months, you spend a lot of time trying to solve it. Drones solve that problem quickly,” he said.