Research on the accuracy of map-aided geolocation on mobile?

General talk
1

Thought this community might know: has anyone out there ever done research on the accuracy of map-aided geolocation on mobile devices?

To explain what I mean: I’ve seen research on the accuracy of mobile devices as GPS receivers, but haven’t been able to find anything that takes into account user correction based on what they see on the map.

For example, when I do mapping, I use the phone geolocation mostly to get to the approximately right area of the map and then position e.g. a new node based on a combination of (in priority order):

  • Existing mapped features, e.g. building outlines
  • Basemap or aerial imagery (with a grain of salt on mobile since aerial imagery may not be offset-adjusted)
  • The phone’s geolocate indicator

My assumption would be that this is much more accurate than the 4.9m accuracy of the average phone GPS on its own with a clear sky, even taking into account the potential accuracy issues with the basemap or aerial imagery used.

The same goes for other apps that collect user-reported locations; even though they would have fewer information layers than an OSM editor, they still have a basemap that a user might use to help position the location they are inputting.

Tried a couple different searches on Google Scholar, but wasn’t able to find anything. Closest was this Masters Thesis from 2015 (PDF) that discusses “GNSS-assisted heads-up” positioning, but doesn’t have any results comparing its accuracy to GPS-only positioning.

2

Do you mean the user manually selects where they are, or the phone takes this into account automatically?

Something that comes to mind is Uber’s satellite ray tracing using 3D building data: https://www.uber.com/en-AU/blog/rethinking-gps/

3

I meant the user manually selects where they are, making corrections to the GPS location based on their understanding of where they are relative to what’s displayed on the basemap.

That Uber article is super cool though, thanks for sharing it!

4

A number of years back, I physically located a few USGS benchmark survey discs, and then compared my phone’s GPS to them (when the phone was physically flat above the disc). Was not able to draw any real conclusions.