How we measure

A claim has to survive a follow-up question.

The numbers on this site are measurement results, not marketing. Here is exactly what they mean, how we measure them and what falls outside them.

0 misreads

What it means

Zero wrongly read labels let through unnoticed. That is the claim, and nothing more. Our vision system reads every label with neural networks trained on your own labels and products, checked by multiple independent engines. Only when engines agree, or a second shot confirms, does a read count as certain. Anything uncertain goes to a human: a case of doubt is not a misread, it's a correctly held read.

How it's measured

Validated on real production runs: every camera read compared afterwards with what was actually on the pallet. Result on those runs: zero wrong labels let through; a share of labels went to the human review queue. We discuss both numbers openly in every conversation.

definition: misread = read wrongly and passed as certain · held for review ≠ misread · read reliability in practice: 99.99%+

10 cm

What it means

With a clear line of sight between tag and receivers, the position on the floor plan typically deviates about 10 centimetres from where it really is. Our acceptance requirement at delivery is wider, below 50 centimetres: among metal, carts and people the position stays usable at shop-floor level. Developed and tested in halls full of metal, machines and obstacles: harsh radio terrain.

How it's measured

At installation we make a 3D scan of the building, import your construction drawings (DXF) and measure the receivers' reference positions with laser precision. Then we compare computed tag positions with physically measured points across the floor. Accuracy applies within the covered area; outside coverage there is no position, no guess. Signal processing uses machine learning to suppress interference from metal and obstacles.

technology: UWB (ultra-wideband) · our own receivers, tags, gateways and firmware

dozens of companies

What it means

Dozens of production companies run on our software and hardware today, in daily operation. Some for many years. Where it fits we start small, with a pilot: so you know where you'll end up before you invest big. What we don't do is tinker for show. Until recently we did all this under the name Blue-bi Systems; the companies and the systems are the same.

references and client names: on request, with the client's consent

on-premise

What it means

The system runs on a server at your site, behind your own firewall. Your data doesn't leave the premises. We have installations at locations without any outbound internet connection: updates arrive physically there, and it works.

hosted works too: the choice is yours, not our revenue model's

Ask us for the proof.

We substantiate every claim on this page with the underlying measurement data, in a conversation or on your floor.

Book a Source Scan

Reinier Weerts

founder, hands-on technologist

r.weerts@unifons.nl

you get a reply from a human: no forms, no ticket numbers