3 ms·

Is the Maps application getting the position data from GPS alone or rather from accelerometer plus earlier good reading (using GPS or cell tower known position)

Is the Maps application getting the position data from GPS alone or rather from accelerometer plus earlier good reading (using GPS or cell tower known position) or some fancy sensor data merging (Kalman filter)?

Bedon292 5y ago
Location data can come from a lot of things that are being pieced together into a really good guess even when parts are missing. GPS and the other satellites like GLONASS. Nearby WiFi hotspots since they are all pretty much known locations. What cell towers you are connected to, which are also known locations, and the signal strength getting you general ranges. Can even be done with bluetooth beacons within building like malls and conference centers I believe. And accelerometer and compass data to fill in gaps too.
- blamazon 5y ago
  In the famous "big dig" highway tunnels in Boston your phone may be connecting to the 850 Bluetooth beacons stuck to the tunnel walls by Waze: https://blog.mass.gov/transportation/massdot-highway/highway-waze-app-tunnel-service-improvements/ https://blog.mass.gov/transportation/massdot-highway/highway...
jfoutz 5y ago
I've only fiddled with this in a hobby way, but it's tricky to get accurate position information by integration of acceleration vectors. you could use the compass to get device orientation relative to motion, you could look for available wifi names to narrow down localization. It's possible to make good guesses, but it's a pain, and they're still guesses. (Well, they were at my level anyway)
- userbinator 5y ago
  They can be quite accurate if well-designed: https://en.wikipedia.org/wiki/Inertial_Navigation_System https://en.wikipedia.org/wiki/Inertial_Navigation_System
- AlotOfReading 5y ago
  The guesses can get scarily good. Typically you start with the GNSS signal as your initial position/velocity estimate and inertial measurements to estimate travel from that point. With decent sensors and no other constraints, you can get lane level accuracy for upwards of a minute at highway speeds without a satellite fix. However, people (and vehicles) don't have arbitrary paths and the places where they do tend to be easy to get a satellite fix. The kinds of places where you have difficult GNSS environments also tend to have networks of paths and streets that can be used to refine your estimate., a technique called "map matching". In good conditions, you can localize position to within a block inside an entire urban area on that alone.
rootusrootus 5y ago
Or if it's attached with CarPlay it will get GPS from the head unit's GPS receiver (assuming there is one, I don't know for sure that all CarPlay head units have their own GPS).

Bedon292 5y ago

Location data can come from a lot of things that are being pieced together into a really good guess even when parts are missing. GPS and the other satellites like GLONASS. Nearby WiFi hotspots since they are all pretty much known locations. What cell towers you are connected to, which are also known locations, and the signal strength getting you general ranges. Can even be done with bluetooth beacons within building like malls and conference centers I believe. And accelerometer and compass data to fill in gaps too.

blamazon 5y ago
In the famous "big dig" highway tunnels in Boston your phone may be connecting to the 850 Bluetooth beacons stuck to the tunnel walls by Waze: https://blog.mass.gov/transportation/massdot-highway/highway-waze-app-tunnel-service-improvements/ https://blog.mass.gov/transportation/massdot-highway/highway...

jfoutz 5y ago

I've only fiddled with this in a hobby way, but it's tricky to get accurate position information by integration of acceleration vectors. you could use the compass to get device orientation relative to motion, you could look for available wifi names to narrow down localization. It's possible to make good guesses, but it's a pain, and they're still guesses. (Well, they were at my level anyway)

userbinator 5y ago
They can be quite accurate if well-designed: https://en.wikipedia.org/wiki/Inertial_Navigation_System https://en.wikipedia.org/wiki/Inertial_Navigation_System
AlotOfReading 5y ago
The guesses can get scarily good. Typically you start with the GNSS signal as your initial position/velocity estimate and inertial measurements to estimate travel from that point. With decent sensors and no other constraints, you can get lane level accuracy for upwards of a minute at highway speeds without a satellite fix. However, people (and vehicles) don't have arbitrary paths and the places where they do tend to be easy to get a satellite fix. The kinds of places where you have difficult GNSS environments also tend to have networks of paths and streets that can be used to refine your estimate., a technique called "map matching". In good conditions, you can localize position to within a block inside an entire urban area on that alone.

rootusrootus 5y ago

Or if it's attached with CarPlay it will get GPS from the head unit's GPS receiver (assuming there is one, I don't know for sure that all CarPlay head units have their own GPS).