Inaccuracies in GPS/maps and impoct on road tolls
Discussion
A contributor said:
I note, after searching a RISKS database of items on "GPS", that a considerable number of observations from writers re GPS errors are actually errors in the mapping data bases that are used in navigation system applications (e.g., automotive navigation), rather than a GPS positioning error due to signal errors per se. This distinction may not be interesting when you are lost in your car, but it is critical in other applications.
GPS position estimates have inherent errors (generally of a couple of meters in "open sky" circumstances, but possibly 100s of meters on some occasions due to "non-line-of sight multipath error" in especially built-up urban areas. Some GPS-Auto-Nav users will have noted temporary errors such as their position being displayed on the wrong road. The difficulty is more subtle than writers surmise. There are indeed errors in the maps being used. Even if a map is correct when installed in your device, roads change.
But at any one moment how can you be sure an error is in the positioning estimate or on the map. You really need to rely on signage if it is available.
But worse than all this is that we are on the cusp of deploying GPS-based road-tolling systems, the majority of which will depend on map-matching algorithms to determine which road you are on or which "cordon" you are in to calculate a charge. These tolling systems will be subject to error for the same fundamental two reasons signal errors and map errors.
The risk here is that tens of companies are building and tens of municipalities and tens of counties are considering investing in GPS-tolling systems that will critically rely on map-matching.
Considering that the very first such system (Germany) cost far in excess of Euro 10^9, these companies, cities and countries are about to put many, many billions at risk. Any decent lawyer could cobble together a class action suit to defeat charges based on map-matching.
StreakyGPS position estimates have inherent errors (generally of a couple of meters in "open sky" circumstances, but possibly 100s of meters on some occasions due to "non-line-of sight multipath error" in especially built-up urban areas. Some GPS-Auto-Nav users will have noted temporary errors such as their position being displayed on the wrong road. The difficulty is more subtle than writers surmise. There are indeed errors in the maps being used. Even if a map is correct when installed in your device, roads change.
But at any one moment how can you be sure an error is in the positioning estimate or on the map. You really need to rely on signage if it is available.
But worse than all this is that we are on the cusp of deploying GPS-based road-tolling systems, the majority of which will depend on map-matching algorithms to determine which road you are on or which "cordon" you are in to calculate a charge. These tolling systems will be subject to error for the same fundamental two reasons signal errors and map errors.
The risk here is that tens of companies are building and tens of municipalities and tens of counties are considering investing in GPS-tolling systems that will critically rely on map-matching.
Considering that the very first such system (Germany) cost far in excess of Euro 10^9, these companies, cities and countries are about to put many, many billions at risk. Any decent lawyer could cobble together a class action suit to defeat charges based on map-matching.
The major source of errors in GPS now that the US has removed Selective Availability has always been the limit of the number of satellites high in the sky as opposed to low on the horizon, and time/clock errors due to multipath.
We were post processing GPS data in 1998 to a positional accuracy of 1-2mm and height/level to 10mm.
We were post processing GPS data in 1998 to a positional accuracy of 1-2mm and height/level to 10mm.
Gassing Station | Speed, Plod & the Law | Top of Page | What's New | My Stuff


