← All projects
RFID + UWB hybrid · WMS / ERP integration

The last scan is not a location.

A window manufacturer's glass panes kept their passive RFID tags; the carts carrying them got UWB. Parent–child pairing gave every pane a precise live location — and the system paid for itself five times over in the first year.

System cost recovered in the first 12 months
RFID + UWB
Pennies on the glass, precision on the carts
WMS / ERP
Locations pushed into the systems staff already use
Dozens
UWB-tagged carts, locating thousands of RFID-tagged panes
The situation

A large window manufacturer tracked every pane of glass with RFID — tagged on arrival, scanned into a storage location. The system was accurate for exactly as long as nothing moved. Then someone shuffled a stack to reach the pane behind it, and the record became fiction. Staff wandered the plant hunting for material, and nobody could time the building of a window.

Accurate until anyone touched anything

Scan-based tracking has a quiet flaw: it records where something was put, not where it is. Panes were shuffled constantly to reach different sizes and materials behind them, so the last-scanned location decayed within hours. Carts of glass relied on RFID portal reads — a gate event, not a position — so a cart was "somewhere past the portal", which is not an answer a worker can walk to.

The cost was constant and mostly invisible: skilled staff walking the plant looking for material, production waiting on glass the building could not find, and no way to measure how long a window actually took to make.

Why this technology

Tagging every pane with UWB would have been absurd — glass panes are numerous, transient and low-margin. Dropping RFID entirely would have thrown away a working identity system and thousands of tags already in circulation.

So the hybrid: every pane keeps its passive RFID tag — identity for pennies, no battery, nothing to maintain. Every cart gets a UWB tag — precise, live location on the thing that actually moves. The expensive technology goes on the dozens of carts, not the thousands of panes.

What we built

The pane inherits the cart

  • A parent–child pairing workflow. The worker scans the cart's barcode, then scans the panes loaded onto it. From that moment each pane's location is the cart's UWB location — live, precise, and updated every time the cart moves, with nobody scanning anything.
  • Located scanners. The RFID scanners themselves carry UWB tags — so every scan event records where the scan happened, to UWB accuracy. When a pane leaves its cart, its last known position is a real position, not a storage-zone guess.
  • Locations pushed into the WMS and ERP. No new screen to learn: the systems staff already used simply started telling the truth about where things were.
  • Carts related to job numbers. As carts moved through the WIP process, each build's current stage and progress became reportable and searchable — which windows are where in production, right now.

Five times the system cost, in year one

The client's own estimate: cost savings of five times the system cost in the first twelve months — search time recovered, production no longer waiting on lost material, and glass that stopped being bought twice.

And a second outcome that compounds: with carts tied to job numbers, the plant could finally time the construction of a window. Quoted lead times became reliable because they were measured, not estimated — which is a customer-service improvement wearing an operations disguise.

The hard part

The hardest part arrived before the hardware did: nobody on the floor believed it would work. Most RTLS projects start exactly there, and the only argument that ever wins is a running system. Once the carts started telling the truth, the sceptics became the heaviest users.

The rest was keeping parent–child honest inside an industrial reality. The client's own discipline did a lot of it — procedures tight enough that an untagged pane was a rarity, and a supply chain worked hard enough that tags were applied at the glass manufacturer, so panes arrived on site already readable by the scanners waiting for them. For the moves nobody scans, RFID portals at key points cross-checked each passing cart against what the system believed was on it, and flagged any discrepancy to staff before it could harden into fiction. And the environment — glass, metal racking, machinery in constant motion — was hostile enough that the UWB vendor developed site-specific antennas to hold accuracy and coverage across the plant.

What transfers
  • Identity and location are different problems — price them differently. Passive RFID answers "which one is this" for pennies. UWB answers "where is it" precisely. Forcing one technology to do both jobs is how WIP tracking gets unaffordable.
  • Track the container, not the contents. The thing that moves is the cart. Let everything on it inherit one good position instead of buying a thousand mediocre ones.
  • Locate the scan, not just the item. A tag on the scanner turns every scan into a positioned event — the cheapest accuracy upgrade in the whole design.
  • Put the answer where people already look. The WMS and ERP were already open on every screen in the plant. Location data that lives there gets used; another dashboard gets ignored.

Planning something like this?

The thinking above is the free part. A discovery call is where it gets specific to your site.

Book a discovery call