In principle, no account data is required for the provision of map and navigation data. Depending on how Mercedes organizes it, Google gets more or less data. If the map data is in the car, Google and Mercedes servers do not need to be consulted to plan the route.
The car branch currently handles merged map data. Routing is done on an external server. The server then sends the relevant map data in small tiles to the vehicle. A connection is required for this and the car will receive an update of the new map piece during the trip. So the route is easy to change, because the external server plans a different route and sends the necessary tiles to the car.
In terms of privacy, this is entirely implementation dependent. If Mercedes sends a unique identifier, or a set of uniquely identifiable parameters to Google (a bit like the idea of ​​fingerprinting a browser), Google can easily connect a person to a car. If the routes the car frequently travels match the phone’s GPS positions, the phone may have been in the car. The owner of the phone can then be connected to the car.
If Mercedes is not identified enough, personal journeys cannot be combined with a unique car. This makes it more difficult to detect, but if the phone matches the coordinates of a Mercedes trip a few times and these trips all leave from the same road, Google will still know that you drive an X Mercedes regularly.
German privacy laws are very strict and we should hope that it is contractually prohibited to use trip data for anything other than navigation. Technically there is no solution to prevent the relationship between a person and the car he drives.