| Versions |
  |
|
| Author |
 Topic  |
|
|
bushido
Belgium
2 Posts |
 Posted - 01 août 2012 : 15:46:21
|
Hi all,
I'm writing an application for android for which I need good accuracy, I use a Galaxy Nexus as test device.
I noticed that, when I draw a car symbol corresponding to my position on a map, the positions don't match with reality, they lag behind considerably.
When my cross a street for example, it will take a few seconds before the car on the map is also crossing that street. It isn't an error in the map data, because when I'm standing still, the car gets plotted on the correct location.
I see exactly the same behavior when I repeat the test with the Maps application on Android.
The position of the car in Google Navigate on the other hand matches reality rather closely.
My question is: how do they do it? Do they use the phone's sensonrs (gyro & accelero), are they interpolating, ... ?
Thanks in advance.
Bushido |
|
|
Ads
|
|
|
gpspassion
93402 Posts |
Posted - 01 août 2012 : 18:06:43
|
| Welcome to the forums...for a very recurring question ever since we got GPS assisted navigation in 2001. The short answer is interpolation, yes, because there is always going to be a one second delay (not to mention the display processing delays) due to the one second update rate of GPS. |
Discounts and Assistance/Réductions et Assistance (Club GpsPasSion) / Où commencer? |
 |
|
|
bushido
Belgium
2 Posts |
Posted - 02 août 2012 : 09:40:53
|
Thanks for the fast answer, although it doesn't exactly answer my question. That's probably because I seem to have kept the question too general and basic.
I wasn't really asking how GPS vendors get round the 1s delay due to the update rate of most GPS chips. I have been working in the industry since 2003 and I know (or at least I think I know) most of that.
The Galaxy Nexus, like the SII & SIII has a SiRFStarIV GSD4t for GPS, which is quite a beast. On top of the GPS update rate delay, it seems to have a variable delay, I imagine because of extensive filtering in the chip. This delay can be as large as 5 seconds when up to speed.
Extrapolating on that kind of data is not a good idea, since the delay is not constant. It's hard to extrapolate if you don't know where to start. There would be side effects (overshoot at extreme angular or linear accelerations), I know these can be mitigated using map data but that has side effects as well ('freezing' at junctions).
I see no overshooting, no freezing, ... in the Google Navigation. A minor remark would be that it always takes it bearing from the road segment when mapmatched, but that's about it.
The only way I can see to achieve that in a structured way is to use the phones sensors (accelero & gyro), apply a kalman filter or a complementary filter on it and correct the GPS data. But that would not be portable to low and mid-end devices, since they don't have a gyro & accelerometer.
So my questions are:
- is the behavior of google navigation the same on high-end and low-end phones ?
- if so, how does google navigation overcome a large & variable delay in the GPS data, without using sensors?
|
|
|
|
gpspassion
93402 Posts |
Posted - 19 sept. 2012 : 18:37:50
|
| I don't think so because I've seen GMN perform similarly on all Android devices I've used, from low (first 7" tablet with no sensors) to high (HTC EVO 3D). I'm a bit surprised you're seeing that much delay, can you try to quantify it ? The only situations where I've seen a lot of delay when plotting raw GPS tracks was in a car with a coated windshield. |
Discounts and Assistance/Réductions et Assistance (Club GpsPasSion) / Où commencer? |
|
|
|
ve7mdt
Canada
170 Posts |
Posted - 02 oct. 2012 : 13:56:22
|
I don't know about Google Map, but for some (not all) consumer street nav software, some attempt to "draw ahead" based on current speed, matching a delay mainly in the app software update (redraw) screen. I did not hear anything about the delay based on the GPS position when driving (may be in a flying rocket??).
Very advanced design such as Alpine over 10 years ago already used gyro and wheel digital pulses for map position alignment, not just GPS.
Another thing I am not sure is if modern day A-GPS has messed up info if those info (such as WiFi) cannot be reliable 100%, but I would assume GPS raw data trumps A-GPS since it is "Assisted" after all.
Don't know much more here. I have Galaxy Nexus and Galaxy S II X. On GPS Test app, S II X showing a lot of triangle sources, but not on GN. Don't know why. |
iPad WiFi w/ RoqyBT to use BT GPSr, LG Optimus G2x, BenQ S6, Samsung Q1, Toshiba e830, Toshiba e805, HTC Advantage X7501, Nextar MN2707 running P7, Magellan Springboard GPS on Visor 2MB, Haicom HI-303III + BT slipper, Holux GPSlim236, eTrex yellow
TomTom 1.9 for iOS, Navigon 2.0 for iOS, Navigon for Android, Garmin Mobile XT, Delorme Street Atlas 2009, Streets & Trips 2010, MapPoint 2010, APRSIS/CE, APRSIS/32, APRSDroid |
|
|
| |
Topic |
|
|
|
| This page was generated in 0,31 seconds. |
 |
|
- 
- 
Accuracy of GPS/Google Navigation ?
