Indoor Positioning

Two vendors can both write "10 cm" and still not be comparable if the measurement conditions differ. Ask: open space or real partitions? Target stationary or moving? Is the published number an average, or a figure met in more than nine cases out of ten? Indoor positioning error fluctuates with crowds, metal structures and obstructions, so a single number detached from venue conditions is hard to use for comparing vendors.

Every step up in accuracy raises hardware, deployment and maintenance costs. Wayfinding, zone-level footfall counts and arrival-triggered notifications are fine at the meter level. Centimeter-level accuracy is normally for collision avoidance on a production line, asset audits or personnel access control, where half a meter of error causes real trouble. Define the use case first and pick the technology second; doing it the other way round leads easily to over-buying.

If the venue already has good WiFi coverage, fingerprinting often needs no new hardware at all. It is worth asking the vendor to run a site survey first and state which equipment can be reused and which has to be added. The answer to this question usually decides the cost bracket of the whole project.

An indoor map is an asset that accumulates over the long term, and it has to be updated whenever floors are rebuilt or units change. Confirm that the venue owns the map data and that it is kept in a common, exportable format rather than a proprietary file locked inside a single vendor's system. For our map data layer we use IMDF, the international indoor map format created by Apple and adopted by OGC as a Community Standard.

Learn about IMDF Map Production

The three technologies carry different maintenance loads. WiFi fingerprinting has to be re-surveyed and recalibrated after AP locations or partitions change, Bluetooth beacons need battery replacement and location audits, and UWB anchors have to stay precisely time-synchronized. Set out the calibration frequency, maintenance scope and response terms in the contract rather than leaving them until after acceptance.

Transport hubs, hospitals and plants each pose their own difficulties: a terminal combines long-span spaces with peak crowds, a hospital has dense partitions and a crowded electromagnetic environment, and a plant suffers blockage from metal structures. Asking which comparable venues a vendor has delivered shows the difference more clearly than comparing specification sheets line by line. Our own work spans all three: indoor map data for the Taoyuan International Airport terminals, station-wide navigation at Taipei Main Station, the emergency department at National Taiwan University Hospital and personnel positioning at the China Steel plant.