By Luis Lopez, AI transportation consultant, CEO of Go Hub.io Holdings Corp and subsidiaries, and host of the Freight Guru Podcast
Counting inventory in a large warehouse is slow, repetitive, and sometimes risky. Someone has to reach high rack locations, scan labels, and compare what is physically there with what the system says should be there. Indoor drones promise to take over a big part of that job. They can, in the right building, for the right kind of inventory. They also have clear limits that are worth understanding before anyone signs a contract.
As an AI transportation consultant, the first thing I look at when a warehouse asks about drones is the labels and the racks, not the drone. If the labels cannot be read reliably, nothing flying past them will fix that.
What inventory drones actually do
An inventory drone is a small aircraft fitted with cameras and, in some designs, barcode scanners or RFID readers. It flies along rack aisles, capturing images of pallet labels and location labels. Software then reads the barcodes, matches each pallet to its location, and compares the results with records in the warehouse management system.
The output is usually a report of exceptions: empty locations that should be full, full locations that should be empty, pallets in the wrong slot, and labels that could not be read. People then review and resolve those exceptions.
Where drones fit best
Cycle counts
Most warehouses do not shut down for a full physical inventory very often. Instead, they count portions of the building on a rolling basis. Drones fit this model well because they can scan many locations quickly and repeat the process frequently. More frequent counts mean errors are caught sooner, before they turn into missed shipments or bad customer data. For background on rolling counts, see our guide to cycle counting and inventory accuracy.
High racks
The upper levels of tall selective racking are where drones offer the clearest advantage. Counting those locations by hand often requires a lift truck, an operator, and a spotter. A drone can read the same labels without putting a person at height or tying up equipment.
Full-pallet storage
Drones work best when each location holds identifiable pallets with a visible label facing the aisle. That is common in bulk and reserve storage. Our overview of warehouse racking and storage types explains which layouts present labels cleanly.
How drones navigate indoors
GPS does not work reliably inside a warehouse. Indoor drones rely on other methods, such as onboard cameras, distance sensors, markers placed in the building, or a tether or ground unit that guides the aircraft. Some systems pair a drone with a ground vehicle that moves along the aisle while the drone scans above it.
Each approach has tradeoffs in setup effort, flexibility, and how well it handles changes to the layout. Ask any provider how the system handles narrow aisles, reflective surfaces, poor lighting, and racks that have been reconfigured since the last flight.
Safety and flight windows
Most operations schedule drone flights when people and lift trucks are not working the aisles, such as after hours, between shifts, or in zones that have been closed off. Mixing small aircraft with active forklift traffic creates collision risk and distractions that most safety teams will not accept.
That scheduling matters. If your building runs around the clock, you need a plan for which aisles are closed when, and for how long. Your safety team should review procedures for emergency landing, battery handling, and what happens if a drone comes down in an aisle.
Indoor use inside a closed building is generally treated differently from outdoor flight. If you are considering outdoor use, such as counting trailers or containers in a yard, confirm the current FAA rules and talk to your own counsel before anything leaves the building.
Battery limits
Small drones have limited flight time per charge. A large building may require many flights, battery swaps, or several aircraft working in sequence. That affects how quickly a full count can be completed and how much staff time goes into managing the fleet. Ask for realistic coverage estimates based on your aisle count and rack height, not a demo in a small test area.
Reconciling with the WMS
A drone scan is only valuable if it connects to your warehouse management system. The integration should pull expected inventory by location, compare it to what was scanned, and present exceptions in a way your inventory control team can work through.
Questions to settle early:
- Does the scan update the WMS automatically, or does a person approve each change?
- How are unreadable labels reported?
- Can the system show the image for each exception so a person can verify it without walking the aisle?
- How long are images and scan records stored?
Label quality makes or breaks the project
Drones read what they can see. Torn, faded, wrinkled, or wrapped-over labels will fail. So will labels placed on the wrong face of the pallet, labels hidden behind stretch wrap glare, or location labels mounted inconsistently across racks. Before any pilot, audit a sample of locations and fix label standards. Our article on freight and pallet labels covers what good labeling looks like.
Drones vs. fixed cameras vs. handhelds
- Drones suit large buildings with tall racks and full-pallet storage, where reaching high locations by hand is slow.
- Fixed cameras mounted on racks, docks, or lift trucks can capture inventory continuously during normal work, without separate flights. They require more hardware and installation.
- Handheld scanners remain the right tool for small buildings, low racks, case or each picking areas, and quick spot checks.
Many operations will end up with a mix. For a broader look at camera-based tools, see our article on computer vision in the warehouse and yard.
What drones cannot do
- Count individual cases or units inside a wrapped pallet.
- Read labels that are missing, damaged, or facing the wrong way.
- Replace the people who investigate and correct discrepancies.
- Fix process problems, such as putaway errors, that cause inaccuracy in the first place.
Worker concerns
Inventory control staff may worry that drones will replace them. In practice, the counting step is what gets automated; the investigating step does not. Be clear about that, involve the team in the pilot, and let them help define what a good exception report looks like. People who understand the inventory are the ones who make the drone data useful.
Running a fair pilot
Choose a zone with tall racks and full pallets. Fix labels first. Measure your current count time and accuracy, then compare against the drone over several weeks. Track how many exceptions were real, how many were false alarms, and how long it took to resolve them. Let those results, not the demo, decide whether to expand.
For more conversations on freight and AI, subscribe to the Freight Guru Podcast.
About the author: Luis Lopez is a Miami-based AI transportation consultant and logistics entrepreneur, the CEO of Go Hub.io Holdings Corp and subsidiaries, and host of the Freight Guru Podcast.