Logistics / Supply ChainVerified Real deployment at a named or precisely described company, with a measured result from a primary source.

Inventory drones count the plant warehouse on a regular schedule, so inventory is right without stopping production (LAPP USA, GKN Automotive, Hyundai Glovis)

Added 2026-10-22 · Last updated 2026-10-26

Manufacturers in this case: Bayer · GKN Automotive · Hyundai Glovis · LAPP USA · all manufacturers using AI

Role

Logistics (plant warehouse: cycle counts and inventory accuracy) · Typical users: warehouse manager, inventory control specialist, logistics or supply chain manager, materials planner, plant controller

The problem

High-bay racks are counted by people on lifts, so full counts happen rarely, often on weekends or by stopping work. Between counts, misplaced pallets and reels, wrong labels and stock that is in the system but not on the shelf delay orders and production. LAPP USA ran only two full warehouse passes a year, with five associates (about 12.5% of the warehouse team) doing weekend cycle counts and still missing targets. GKN Automotive's inventory specialist checked locations from an aerial lift, and twice a year a full inventory needed 5 aerial lifts and 10 operators; the company wanted regular stock takes without stopping production for a whole day.

How AI helps

Indoor drones fly the aisles on a schedule (LAPP's Corvus drones fly nightly missions after people leave, with no human pilot), and read pallet and rack labels with cameras and computer vision, without GPS or changes to the racks. The set-ups differ: EYESEE, for example, locates the drone with two mobile ultra-wideband (UWB) tags placed on the ground during each flight, with no fixed infrastructure. Each scan is compared with the warehouse management system (WMS) and the drone keeps a photo of every location, so the team gets a daily discrepancy list (missing, misplaced, mislabelled) and corrects it from a desk instead of walking the racks. People still decide each correction in the WMS.

Step by step

1. Record the baseline: full counts per year, hours and people on cycle counts, overtime, inventory accuracy by location, and orders or production stops caused by stock not found.
2. Check the basics drones depend on: one readable label per pallet or location, consistent label height and position, and lighting in the aisles.
3. Start with the high-bay reserve area that is hardest to count by hand, and schedule flights when the aisles are empty, as LAPP does at night.
4. Connect the scan results to the WMS so the team gets a daily list of discrepancies with photos, and agree who corrects what and by when.
5. Use the discrepancy patterns to fix causes (labelling, put-away errors), not only the stock records.
6. After three months, compare accuracy, count frequency, labor hours and overtime with the baseline, then decide whether to extend to other areas or sites.

Tools

• Corvus Robotics Corvus One (autonomous inventory drones and AIMS platform): used at LAPP USA
• EYESEE drone inventory (invented by Hardis Group, marketed since March 2022 by Darwin Drones): used at GKN Automotive and Bayer
• Hyundai Glovis in-house drone system linked to the site's consolidation-center system: used at HMGMA
• Other tools to evaluate: Gather AI, Verity, and fixed or forklift-mounted camera systems
• Any LLM (ChatGPT, Claude, Copilot, Gemini) with the prompt below, on a discrepancy export, to sort and explain what to fix first

In the AI tools index: Corvus Robotics · EYESEE (Hardis Group) · all AI tools used by manufacturers

As reported by the source; most figures are vendor or self-reported and not independently verified.

Human review: cautions

• Drones need readable labels and clear aisles; poor or inconsistent labelling limits what they can read, as LAPP found and fixed.
• The drone finds discrepancies; people still decide and post every correction in the WMS, and the cause (put-away, labelling) has to be fixed too.
• Check safety rules for flying near people and forklifts, and whether flights must happen outside working hours.
• Figures are published by the vendors (Corvus, EYESEE) and, for Hyundai Glovis, reported by the press from the company's statement.

You are an experienced inventory control analyst in a plant warehouse.

I will paste a cycle-count or drone-scan discrepancy export [location, expected item and quantity from the WMS, item and quantity found or label read, date of last movement, photo reference if any], and optionally our item master [item, unit cost, A/B/C class, used in production yes/no].

Do this:
1. Classify each line: missing (in WMS, not found), unexpected (found, not in WMS), wrong location, wrong quantity, unreadable label.
2. Rank the lines to fix first: items used in production or on open orders, A-class items, then the rest; give the reason for each rank. Copy quantities and costs as given; do not compute totals yourself, and give me spreadsheet formulas instead.
3. For wrong-location lines, look for matching items elsewhere in the list (the 'missing' pallet may be the 'unexpected' one in another aisle) and list possible matches.
4. List patterns that point to a cause: the same aisle, the same put-away time or shift, unreadable labels in one zone.
5. Draft the correction list for the inventory team, with a column for the person who checks and confirms each one.

No WMS change is made without a person confirming it on the shelf or on the photo.
Use only the data I paste in; if something is missing, write 'not found' or list it as a question instead of guessing. Do not send, order, post or change anything in any system: produce drafts for my review and approval.

Related cases

• AI orchestrates a plant-attached warehouse: more loads shipped straight from the plant, fewer shuttle moves (P&G Lima)
• AI reads supplier delivery notes and prepares the goods receipt in the ERP (Siemens, RAUMEDIC)
• AI replenishment recommendations so reorders are not missed
• AI moves fast-selling items closer to the pickers and shortens pick routes in the warehouse (Coram, Gildemeister)

All 12 Logistics / Supply Chain use cases →

Sources

• https://www.corvus-robotics.com/case-study-lapp
• https://www.eyesee-drone.com/client/gkn-automotive/
• https://www.eyesee-drone.com/client/bayer/
• https://www.ajupress.com/view/20250707152254306
• https://www.eyesee-drone.com/the-eyesee-inventory-drone-plots-a-new-course/