Perfect in the Demo — The Real Reasons RTLS Accuracy Falls Apart in the Field (The Science of Accuracy, Part 1)

For most teams evaluating a real-time location system (RTLS), the first encounter is a demo. You walk across the demo room holding a tag, and the dot on the screen follows you smoothly, accurate to a few tens of centimeters. It is convincing. Then the contract is signed, the system is installed at the actual site, and something strange happens. Some zones are just as accurate as the demo — but at the end of a corridor or in a corner, the error grows several times larger. Rebooting the hardware changes nothing. Neither does a firmware update.

Is the equipment defective? Usually not. This article breaks that gap down into three reasons. The first two are widely known; the third is one that even much of the industry rarely addresses properly.

Ⅰ. The culprit is the space, not the equipment

From a radio wave's point of view, a demo room and a real site are two entirely different worlds. The demo room has no obstacles, and the anchors — the fixed devices that receive positioning signals — sit in ideal spots. A real factory or warehouse is full of steel structures, metal equipment, forklifts, and stacked goods whose position changes every day.

Same hardware, same algorithm — but a different space means different performance. That is why the true measure of a positioning system is not "how many centimeters in the lab" but "how little its performance collapses in an unfamiliar space."

Ⅱ. The first troublemaker — echoes (multipath)

Sing in a tiled bathroom and your voice bounces off the walls, coming back layered over itself. Radio waves do exactly the same. The signal a tag transmits travels straight to the anchor, but it also produces copies that reflect off metal walls and equipment and arrive a moment later. This is multipath.

Positioning works by timing a signal's arrival to the nanosecond to calculate distance — and when the original and its echoes overlap, that timing blurs. The reason UWB (ultra-wideband) is stronger at positioning than other radio technologies is that it transmits extremely short pulses, which makes the original easy to tell apart from its echoes. Yet even UWB is not entirely immune to multipath in metal-heavy industrial environments.

Ⅲ. The second troublemaker — blockage (NLOS)

When someone calls you from behind a wall, you hear the voice, but the direction and distance stay vague. Radio waves are no different. When stacked goods or equipment block the path between anchor and tag, the signal arrives by a detour — and that detour makes the measured distance longer than the real one. This is the NLOS (Non-Line-of-Sight) problem.

What makes NLOS particularly tricky in industrial settings is that the site changes every day. A sight line that is clear today gets blocked tomorrow by newly delivered material. A layout that was perfect at installation may no longer be a month later.

Ⅳ. The third and quietest troublemaker — angles

Everything up to this point is fairly well known. What is far less known is this: accuracy can differ from zone to zone even when there are no obstacles at all.

The cause is the directions from which the anchors see the tag — in other words, geometry. In the middle of a hall, where anchors are visible evenly in every direction, positions come out sharp. At the end of a corridor or in a corner, where the visible anchors cluster toward one side, positions blur even at identical signal quality. Every signal is received perfectly well — and it still happens.

The single number that captures this effect is GDOP (Geometric Dilution of Precision), a concept that comes from GPS. We will unpack exactly how it works in the next installment, using nothing more than two fingers. For now, remember one thing: the first two troublemakers are properties of the site environment and can never be fully removed, but the third is a problem you can measure and handle by design. And whether a system handles it or not is what separates one level of RTLS from another.

Ⅴ. So what should you ask before deploying?

Anyone can answer "How accurate is it, in centimeters?" with a lab number. The better question is this: "Can you show me where in our facility the system will be accurate — and where it will be weak?"

To answer that question, ORBRO starts by drawing an error map of the space. We measure the geometric conditions zone by zone to establish a baseline, then apply different positioning strategies to different zones on top of it. It is a subject ORBRO has worked on long enough to have documented in a technical white paper.

Closing

Positioning accuracy is decided not by a number in a catalog but by the spatial structure of the site. "The Science of Accuracy" series unpacks this subject in plain language over four installments. Next up is the third troublemaker from this article: how GDOP works.

If you are considering an RTLS deployment, you can start with a structural diagnosis of your site. Contact ORBRO to get started.