[Software] Google uses 3D mapping to improve GPS accuracy in high-density urban areas

[Dark]

Google has used 3D mapping in conjunction with machine learning to improve the accuracy of geolocation systems in densely populated urban areas.

   In densely populated urban settings, geolocation systems can show inaccuracy, for example by locating the wrong side of a street or even in the wrong building.

   It is the "last great unsolved problem" of GPS, as indicated by Google. This is due to how the signals of these systems are reflected in large cities - large buildings block the direct signals sent by satellites - and therefore they have decided to use 3D mapping to correct inaccuracies.

   The solution they use at Google includes 3D models of buildings, raw GPS measurements and machine learning, as explained in the official Android Developers blog.

   The module on Google Play that contains the 3D mapping for corrections has models of buildings from more than 3,850 cities around the world. When the user uses the GPS service on his mobile, the device downloads the three-dimensional models of the city where he is, to make corrections for that specific city with the help of algorithms.

   It is a solution compatible with GPS, but also with other satellite navigation systems, such as GLONASS, Galileo, BeiDou and QZSS.

   This solution has now been introduced on the Pixel 5 and Pixel 4a 5G phones with the December update, and works in conjunction with Qualcomm's Snapdragon 5G processor that integrates the two devices to improve accuracy in cities.

   The company has reported that this solution will be available for the Android ecosystem (starting with Android 8.0) early next year.