Tracking With Markers
If you’ve seen any of our videos, you might have noticed that ARDI-AR uses markers to identify your equipment. You might have also seen examples of AR working without markers in some applications and in promotional videos.
Markerless tracking is possible, but our testing has shown it’s largely unsuitable for industrial environments, for several reasons.
Most important among these is the fact that ARDI needs to know not only what an object is, but which object it is, which makes markerless tracking significantly harder.
GPS Accuracy
Truly markerless tracking depends on an extremely accurate GPS or other positioning signal.
Unfortunately, industrial environments tend to be extremely bad for this, as they are largely indoor and surrounded by metal – this severely degrades your GPS accuracy, meaning that you we can’t determine your exact location or confirm which asset you are currently viewing.
A lot of demonstrations for this style of AR will be in clear-sky environments – a situation that none of our current users have been lucky enough to have.
Visual Uniqueness
The alternative form of markerless tracking uses the equipment itself as a marker. By analysing the unique patterns of light and dark on an item, it can be recognised by the camera. However, the problem here is that these patterns need to be unique.
The object has to be completely visually distinct from any other on-site. This is very rarely happens on an industrial site. One pressure transmitter looks like any other transmitter, which in turn may look a lot like the head of a magnetic flowmeter.
And due to the GPS accuracy problem above, we can’t utilise GPS assistance to help narrow down which of the assets we are viewing.
Asset Count
Searching for visually unique assets is an computationally expensive task. While a modern smartphone might be capable of searching for 20 or even 100 different items, searching for the thousands of possibilities on an average site would cause your average smart device to grind to a halt – it’s simply requires too much computing power dealing with so many possibilities.
ARUCO markers such as the ones ARDI-AR use are inexpensive to search for, since they all share the same basic outline. ARDI-AR simply searches for the box, and then reads the code from inside it. This means that – unlike markerless tracking – we can scan for large numbers of different devices with only a small amount of additional computing power.
Distance & Reliability
Markerless tracking has a distinct problem with distance. Like scanning a barcode, the object usually has to be taking up a reasonable percentage of your screen in order to be properly detected, and it can only be picked up from the ‘front’.
Using markers, you can set up additional scanning points at all four sides of an asset, and scan them from a significantly greater distance. The high contrast and regular shape of a barcode-style AR marker means that the markers are scanned more consistently, tracking is more stable and the distance you can read it from is much higher.
However, we do have web-based addons that use alternative marker technology, such as **barcode**, **QR code** and **NFC tags**. These are designed for users who already have a marker system in place and don’t want to add additional markers to their equipment. Although the read distance is radically reduced compared to AR markers, you’re able to utilise your existing asset identification system and still get almost all of the advantages of ARDI.