5 mm Indoor positioning system accuracy


We are developing a system in which we required indoor positioning accuracy of 5 mm, can it be achieved using Decawave system?

Yes, but…

That level of accuracy is in a zone that requires more than ordinary efforts. We have built such systems, notably a basketball arena with 54 anchors which resulted in precision of a few mm. Our location algorithm is adept at taking in large amounts of data and producing the best location output.

We need to distinguish between accuracy and precision. Precision, or repeatability, is the stability of a measured point in space. It may have an average offset to the true position. Accuracy is the error to the true position.

If you have a precise system, you can generally convert it to be accurate by applying a correction mapping. That is, you calibrate the system for the systemic errors. This can be time consuming to do, but it is one way to convert precision into accuracy.

Many applications don’t require accuracy per se, but do require precision, in which case you have it somewhat easier.

You will need a fair number of anchors. The more the better. The system can be made more and more accurate without theoretical bound by adding more anchors, but eventually it gets ridiculous at some point.

Averaging helps. That is, you want a system where the tag is located at some multiple of the desired rate and the output is averaged, or better yet, a smoothing filter, to produce a more stable result. This has the effect of multiplying your anchor array density, sort of. Our system supports ~4000 locates per second, which often gives capacity to run tags at a high rate to allow smoothing.

Clear line of site and good anchor array geometry will be important. A basketball arena with anchors in the ceiling structure is about as good as it gets.

Z axis is always worse. Just the nature of it. The XY axis is better because there are anchors in both directions. There are generally no anchors below the floor, so Z axis is one sided sensing. If your tags are constrained in some way (say top of a machine) where the Z height is known, that can help a great deal and the fixed height serves as a virtual anchor in the solution.

Survey is critical. We surveyed the arena with precision laser surveying equipment (total station). You need anchor locations well within 1 cm to do well and the total station can locate them within 1 mm generally. Poor survey is the number one cause of bad output results in our experience.

Tag rotation introduces error. The nature of all UWB antennas is that they vary some amount in antenna delay depending on direction. The best ones are under 2 cm, many are over 10 cm. Since this is larger than your desired precision, a tag whose orientation changes will introduce location errors as it appears closer to some anchors and further from others due to rotation. If your tag is fixed orientation (say mounted to top of machine), that will help a lot.

Doing a 5 mm system is PhD level UWB deployment. The Decawave UWB technology can do it, but a novice is unlikely to achieve it, and it won’t be cheap or easy.

Mike Ciholas, President, Ciholas, Inc
3700 Bell Road, Newburgh, IN 47630 USA
+1 812 962 9408