- The solution from technology company Verity was tested in a California warehouse to track to location of On-branded shoes stored onsite
- Verity’s drones use an AI system to plan their route, travel through the facility pinpointing the location of RFID tagged shoes, locate any obstacles, and conduct recharging as needed
Logistics and supply chain company Maersk has gained more efficient inventory tracking with a technology pilot of an RFID-based, drone solution. The results of the three-month pilot conducted last summer found that the company could locate On shoes in its California warehouse with drones autonomously scanning as they flew past products.
The solution was provided by Verity, an AI-inventory management and drone-based technology company that already offers computer vision, inventory tracking solutions. Recently, the company has integrated RFID reader units and antennas into its drones so that they can not only identify products that can be visually identified, but goods inside boxes based on interrogations of the products’ UHF RAIN RFID tags.
The technology piloted suits Maersk’s approach of deploying cutting edge solutions to solve problems. Predicting that the technology can enhance the company’s logistics operations, “(we were) was thrilled to participate in this pilot program due to the advanced capabilities that drones offer for inventory management,” said Jason Walker, head of Maersk Contract Logistics in North America.
Piloting Autonomous Inventory Management
The pilot was conducted at Maersk’s high-volume, fast-paced warehouse environment. It was intended to track On shoes, which are already tagged with passive UHF RAIN RFID tags for inventory and supply chain management. Each tag (one tag for each pair of shoes) is encoded with a unique ID that links to that product’s identity.
The system is able to detect tags at an estimated reading speed of up to 1,000 tags per second, said Markus Hehn, Verity’s cofounder and CTO. “This allows for precise identification of each item with a 99.9 percent success rate. The technology enables Verity to precisely pinpoint location of each tag in the warehouse,” based on algorithms built into the system, said Hehn.
With the technology pilot, a group of drones, with two antennas in front, were set up to traverse the area of the warehouse where the On branded shoes were located. They mapped out their own routes, creating a digital twin of the warehouse area, and travelled the aisles, capturing tag reads as they went. They used onboard intelligence to slow, pivot or rotate to ensure detailed and accurate tag reads as well as using existing computer vision technology to avoid any obstacles in the area.
Considering the Next Step
Through this pilot, said Walker, “we observed substantial benefits from RFID-enabled drones, such as improved inventory accuracy, more streamlined operations, and better resource allocation. More efficient inventory capabilities allow our operational teams to focus on more strategic initiatives, ultimately leading to greater efficiency and service excellence.”
“Given the highly positive results, we are actively exploring opportunities to further integrate this technology and assess its potential for broader implementation across our operations,” said Walker.
A History Innovating Drone Technology
Verity, with a decade of experience in Switzerland and deployments in 15 countries, has been providing autonomous drones for inventory control for the past five years. The company developed drones intended to operate in an indoor environment, with software accessible from web interfaces to instruct the drones, and to manage the drone collected data about what they found in the warehouses where they scanned inventory levels.
“When we perform inventory control with our existing system, a good proxy for what we can do is whatever a person can do without touching the goods. A drone can do that by looking at it with a camera,” said Hehn. The resulting computer vision system provides an update of stock levels in the warehouse.
“We know that there’s certain limitations to that and we know that that’s one of the reasons people like RFID—you can see inside the box,” he said.
RFID Reading Integrated into Drone
While adding off-the-shelf RFID readers to a drone can create excessive weight, Verity opted to build the components directly into their own drones.
“We don’t have to attach a whole reader we can just take basically the chips and the antennas and everything and fit it into our existing electronics,” Hehn said. There is no need for a separate battery, display or buttons. As a result, the drone expends more power than it would without the RFID reading, but less than 10 percent more.
Recharging the drones takes place automatically. One of the advantages of tracking inventory with drones is the system can be scalable depending on the size of the warehouse and amount of inventory to be tracked.
“Basically, our drones know how long they can operate and can plan their routes accordingly. Then they automatically come back to a charger, recharge and then fly out again,” he said. “You can just add more drones as you need them.”
Behind the scenes, an orchestrator feature in the software tracks the movement of drones, and provides relevant instructions to the drones based on inventory that needs to be counted.
Challenges in Modern Warehouse
Inventory counting in today’s warehouses is more challenging than ever. Warehouses have high product density, which means high RFID tag density as well. Subsequently, companies often use either handheld scanners with relatively limited range or they use RFID tunnels as goods are received and then shipped out.
The drone is designed to read these tags in high volume wherever they are stored. And they also must identify specific locations. That required some engineering on the part of Verity.
“For us to manage this onslaught of RFID tags and to be able to distinguish what’s going on here, and which tags are where, was probably one of the most interesting technical challenges,” said Hehn.
As the drone is picking up many tags at once, it uses algorithms based on signal strength or angle of arrival. Machine learning and AI further identifies the location of the tag, and links it to the known location of the drone based on Verity’s localization technology. To refine those results, the company implemented its own methods for localizing where the tags are, leveraging the fact that the drone is flexible and mobile and can rotate to change the orientation of its antennas.
“It’s a little bit like how a human can move a handheld RFID reader around but in a much more deterministic fashion,” Hehn said.
The drone can adjust the settings of the RFID reader accordingly if many tags are responding as well as the power settings to read all of those tags that respond most reliably before moving on to the next location. It can also slow down the RF responses from the tags to capture each unique ID in an orderly format “so that you have more chance to hear from all of them,” added Hehn.
Reading Tags Outside of Business Operations
The experimental phase of the pilot ran for several months “and in that time we ran two generations of drones: a first generation for a month, then a generation benefitting from hardware improvements, and then ran the second set of drones for the next month,” Hehn said.
Maersk management chose to run the drones whenever it was convenient because of low activity on the warehouse. They could use a dashboard in the web interface to instruct or set schedules for drones to fly.
“They can also mark which areas the drones are allowed to enter or not enter,” he said. For example, if maintenance is underway on one aisle, they could block out that aisle so that the drones didn’t go there.
Safety Measures Built In
These drones were still equipped with computer vision plus RFID so whenever they were flying around they were collecting RFID measurements as well as camera images to see and understand the environment. Verity and Maersk could then correlate those results to see that’s whether computer vision results matched RFID scans.
Computer vision helped drones avoid obstacles. In an extreme case, if the drone was completely obstructed at the end of an aisle, it could wait until eventually the battery ran out and it could then perform a “soft stop,” which means that it would go to the ground and shut itself down.
The drones could fly for around 20 minutes but they don’t typically operate until the power is almost out. As one goes back to charge, another one takes over that aisle and continues to work.
“They know how to best make use of their battery in terms of how far to fly and how quickly to recharge,” said Hehn.
More Frequent Inventories
According to Verity, one of the biggest benefits is to provide better visibility into inventory levels so that a company can be sure that if they need to ship something out, it’s actually onsite.
While all companies check their inventory levels in one way or another, those adopting Verity’s drones would be able to check their inventory more often than they would otherwise do and therefore have much higher confidence in the goods that they have being right, Hehn said.
And with good inventory data, companies are less likely to manage safety stock to prevent out-of-stock situations. That added inventory can pollute a warehouse with more goods than are needed, which means that forklift drivers end up wasting time because they can’t find the things that they need.
Eliminating Pallet Hunters
In fact, some warehouses have teams of staff known as pallet hunters that try to find a pallet that is not where it’s supposed to be. Hehn argued that creates a lot of operational inefficiencies.
Some pallet hunters may be driving around a large warehouse looking for particular goods or even using binoculars trying to find the palette that they’ve lost. “Doing that in a facility of 1,000,000 square feet it’s not exactly fun,” he said.
As more goods are tagged with RFID in the coming years, Verity expects its RFID solution to be leveraged in greater volume. It already serves companies that handle apparel goods (which are more commonly tagged) but also those in food and beverage, general merchandise, furniture and automotive spare parts, with computer vision drones.
For those that have some RFID tagged goods, but other goods that are not tagged, the Verity drones continue to use both RFID and computer vision technology in conjunction. “If your goods aren’t tagged we can help you a lot using our existing camera technologies. if your goods are tagged with RFID we can do even more using RFID,” said Hehn.
Potential Growth
The software creates a full digital twin of the facilities so that a 3D map may contain all of the racking based on the company’s naming schemes, “so when our client’s warehouse management system tells us that they’ve got a pallet of 100 shoes stored in storage location A58B, our system knows in full 3D space exactly where that is meant to be as well as how the drone can get there.”
The software does allow users to request an audit of specific locations, for either a full scan or a scan of all of the goods that have been touched during the last day.
Verity is in discussion with some of its existing customers that use drones for inventory management to deploy RFID.
In the meantime, at Maersk, said Walker, by incorporating RFID tracking, for tagged goods, “we expected to significantly improve our inventory tracking and management with greater accuracy and efficiency. This technology has the potential to streamline our operations and enhance the level of service we provide to our customers.”