NFI Industries’ inventory drone deployment takes flight with Gather AI

Founded in 1932, NFI Industries is one of the oldest and largest privately held and family-owned third-party logistics providers in North America. It serves a variety of segments, including food and beverage, retail and e-commerce, manufacturing, and consumer packaged goods.

The Camden, N.J.-based company offers everything from transportation, distribution, brokerage, port drayage, intermodal, global logistics, and real estate. To accommodate the ever-evolving needs of its customers, NFI continues to grow its warehouses, operating more than 50 million sq. ft. of space.

However, scaling operations come with its own set of obstacles, such as finding skilled workers. There’s also the issue of efficiency, as many of NFI’s customers have contractual key performance indicators (KPIs) around inventory accuracy.

In addition, the high cost of real estate has forced NFI to add more racking to maximize the use of the space. Doing more within the facility makes logistical sense, but it can present a safety hazard for individuals conducting inventory cycle counts on lifts.

One way to solve these problems is to add more automation to the warehouse. Of course, autonomous mobile robots (AMRs) are great on the floor, but NFI needed something in the air.

Drones, otherwise known as uncrewed aerial vehicles (UAVs) have traditionally been relegated to outdoor activities in the agriculture, construction, utilities, and energy industries. They typically use GPS to navigate, but advancements in image recognition technologies and artificial intelligence are bringing these flying robots inside the four walls of the warehouse to provide automated inventory tracking and management.

NFI deploys warehouse drones

At NFI, the manual process of inventory management consists of scanning each location of the rack with an RF (radio frequency) barcode reader to capture the information in the warehouse management system (WMS). It is a time-consuming task that is inefficient, with warehouse employees averaging about 40 pallet locations per hour. The flexibility of drones was an appealing option.

In 2023, NFI and Gather AI, a provider of AI-powered inventory management systems, put together a proof of concept with one of NFI’s retail customers. Their goal was to use drones to increase cycle count times and to reallocate labor to other areas of the warehouse.

The scope of the project started with one aisle, where productivity, safety, intervention rate, accuracy, and indirect time were tracked daily for over two months.

“Coming out of that, we did have really good results,” said Cody Merritt, director of solution design and innovation at NFI Industries, during a presentation at MODEX 2024 in Atlanta. “Our productivity was five times faster than traditional methods, and this included indirect time of using the drones or getting them set up in the aisle.”

To that end, Merritt and the team said they expected the drones to malfunction at some level, but that never happened, so there was no need for any interventions with the equipment. “They flew autonomously, and they came back. So that was something we stopped tracking at the beginning,” he said.

From a safety standpoint, it kept the employees safely on the ground rather than on lifts, plus they enjoyed working with the drones, which increased job satisfaction. As far as inventory management, in Month 2 of tracking, the testers reported 99.98% accuracy for scanning license plate number (LPN) labels.

“We could rely on the drone to get us the data that we were looking for,” Merritt said.

Gather AI monitors inventory

Using the technology was straightforward because Gather AI handles installation and creates a digital map of the warehouse. No additional lighting, power drops, Wi-Fi, or beacons were needed, and the system works with any WMS, as well as existing labeling and workflows.

Once the setup is complete, an operator with an iPad app can simply choose the area of the warehouse to be scanned, place the drone on the ground, and press takeoff. The drone flies autonomously — no pilot required — using labels placed on the racks to navigate and take pictures of the bins selected.

One drone can scan 300 pallets per hour, but an individual can operate three drones simultaneously to scan 900 pallets per hour, claimed Gather AI. The drone, which is off-the-shelf hardware, can fly without modifications and land itself when it’s done collecting data or when the battery is low and needs to be swapped.

After the drone mission is complete, images are uploaded to the cloud for processing by a machine learning algorithm and displayed on a dashboard that operators can use for inventory matches, to identify empty bins, or to search for missing LPNs.

“It’s going through and determining what is there in real time, and then comparing it directly with a warehouse management system,” said Sean Mitchell, Gather AI’s vice president of customer success. “So anytime in which things do not match, we’re raising an exception.”

The company said its technology can help find things faster, improve productivity, and provide engineers and operators with more information.

For NFI, for example, instead of having someone conducting a cycle count using a barcode scanner on about 40 to 60 pallet locations per hour, that individual can operate three drones scanning up to 900 locations per hour.

“That means you’re taking your resources and turning them into something that is 15 times more efficient,” said Mitchell. “And in doing so, you’ve got full traceability, you have a picture of every location. You can also determine, using AI, whether something is located there or whether it’s empty.”

“We can also do estimations when it comes to case counting and occupancy of every location,” he said. “And everything is visual. It’s no longer just lines on the spreadsheet.”

Merritt confirmed that Gather AI was a seamless implementation: “It was quick. Within a couple of weeks, we’ve got the building set up with the labels and we’re flying.”

“It became a nice tool to our kit of other technologies that we can apply to other customers quickly,” Merritt said. “And there’s no integration with WMSes, which can be cumbersome.”

Drones have an unexpected upside

Following the successful pilot project, NFI mapped the Gather AI across the entire building, using one to two drones. The next phase is scaling this to its West Coast facilities, which is more of a campus environment for multiple clients.

NFI’s long-term goal is to be able to deploy drones in any of its sites in North America, based on customer requirements. As more locations are added, the company will also look at how accurate inventory cycle counting affects outbound productivity.

“That’s something that, as we work more with this ops team on the West Coast and other sites, we’re excited to see how that adds into our payback in our business case,” Merritt said.

One area that can’t necessarily be measured but is important to the success of the project is how it impacts people. In the beginning, NFI employees resisted the drone deployment, believing it could take their jobs. But by the second month of the deployment, there was an attitude change.

“They were excited about it,” recalled Merritt. “They got to go home and tell their families about using the drone or working in another area of the warehouse. And, you know, it became a win.”