Google- Location determination with 3D correction from Android 8.0

 

Pixel 5 - Yellow: Route, Red: Without 3D correction, Blue: With 3D correction
Google 

The improvements introduced with the feature drop for pixel smartphones around the more exact location determination are to be rolled out at the beginning of the coming year on all smartphones starting from Android 8.0 Oreo over the Google Play services. Google improves GPS accuracy, especially in large cities with dense development.


Reflections lead to shifts


Anyone who has ever been on foot or by car in the streets of a densely built city might know the problem. In Google Maps, the user’s location is displayed incorrectly, for example, on the wrong side of the road or moved around a block. The cause: Global navigation satellite systems (GNSS) assume that they determine the position by means of trilateration with a direct line of sight to the user. However, the steel and glass fronts of high buildings may cause the signal to be blocked or reflected and to travel a longer distance, which may lead to: that the location is moved by a few or the same several dozen meters in apps like Google Maps.

In Google Maps the location can therefore be wrong
In Google Maps the location can therefore be wrong


As a map provider, Google has access to an extensive archive of 3D data for numerous buildings in major cities and uses this information to correct the location. As part of the Google Play Services, Google offers the 3D mapping aided corrections module, which comprises the 3D data of buildings from over 3,850 cities worldwide, divided into tiles. The function is only triggered if it has been detected via the Activity Recognition API that the user is on the move as a pedestrian. If this is the case, the 3D data with a size of about 20 MB is loaded onto the smartphone and used to correct the positioning. In the future, Google also wants to offer the correction when driving.


Fused location provider receives corrected data


Within the module, Google also wants to have solved the "hen-egg problem". If the location was not determined correctly, how should the system know which buildings reflect or block the signals? The company does not go any further, but explains that large-scale corrections are only feasible for Google, as they have access to 3D data and GPS raw data and use machine learning.


The corrected location data will be accessible via an API to the Fused Location Provider (FLP) on the apps. Under Android, the FLP is the central collection point for all information on location determination. Since Google also uses information such as known WLANs, the location determination does not necessarily have to be done via GNSS. Via another API, Google also shares the information with the GNSS chip so that the next point can be calculated more accurately.


Improvement by 75 percent from Android 8.0


With the current feature drop, Google released version 2 of the 3D-supported correction, which aims to reduce the probability of the location being determined for the wrong road side by 75 percent on pixels 5 and pixels 4a 5G. All other smartphones still use version 1 of the solution, which should only reduce the probability by 50 percent. Version 2 will be available for all smartphones starting with Android 8.0 at the beginning of next year. The update is done via Google Play Services and works with GNSS chips from Broadcom, MediaTek and Qualcomm.


The solution is also not limited to the USA’s skyscraping cities. Google lists over 3,850 major cities worldwide for which 3D data of buildings are available. These include all major cities in the USA, Canada and Mexico, all major cities in Europe with the exception of Russia and Ukraine, all major cities in Japan and Taiwan and all major cities in Brazil, Argentina, Australia, New Zealand and South Africa. As soon as new 3D data for Google Maps are added, these should also be used for the correction of the location determination and are available in the FLP. The system works with the GPS of the USA, GLONASS of the Russians, Galileo from Europe, BeiDou from China and the Japanese QZSS.

Next Post Previous Post
No Comment
Add Comment
comment url