55

I thought I have read somewhere that the accuracy of Android's GPS is about 10 cm?

Can anyone verify or correct this? Reason being is that my application I am trying to develop keeps track of locations a user has visited and it would help me out tremendously to know the margin of error I can expect.

starball
  • 20,030
  • 7
  • 43
  • 238
Anthony Forloney
  • 90,123
  • 14
  • 117
  • 115
  • 2
    Maybe you will be interested in this: http://gis.stackexchange.com/a/43657/22469 – Áxel Costas Pena Sep 30 '13 at 10:58
  • [Deep Dive into Location by Reto Meier](http://android-developers.blogspot.co.uk/2011/06/deep-dive-into-location.html) might be useful. [GIS stackexchange QnA](http://gis.stackexchange.com/questions/43617/what-is-the-maximum-theoretical-accuracy-of-gps/43657#43657) – GuruM Dec 17 '13 at 05:39

5 Answers5

56

10 centimeters? No chance. In any event, Android is just a device OS; the actual accuracy of a GPS device is dependent on the device's chipset. Android may be theoretically capable of accomodating devices with that level of accuracy, but that's it.

The accuracy of GPS devices is normally not presented as a simple distance, anyway. Usually, a device's accuracy is given as a distance along with the percentage of measurements that are within that distance from the true location. A typical accuracy for a hand-held GPS device would be something like 30% of your measurements within 50 meters of the true position (yes, the devices are typically that inaccurate).

MusiGenesis
  • 74,184
  • 40
  • 190
  • 334
23

Depends on the actual GPS hardware but it is reasonable to expect about 10m horizontal accuracy now that SA is turned off (it was 100m when SA was on). Vertical accuracy is usually about 3 times worse than horizontal.

If the GPS hardware can receive WAAS correction it can do better, about 7.6m worst case, 1-2m best case. Most phones don't have WAAS though (do any?).

There is a table of expected accuracy for various kinds of devices in the WAAS entry of Wikipedia.

progrmr
  • 75,956
  • 16
  • 112
  • 147
  • 1
    You'll notice in that table that the phrase "excluding receiver errors" is added for the measured GPS and measured WAAS values. This is funny, because receiver errors in handheld GPS devices are *massive*. – MusiGenesis Mar 15 '11 at 15:19
20

Drop the "c" and you've got it about right. Professional grade receivers such as Trimble's Nomad series can do 2-5m accuracy before post-processing, and 1-3m afterward, which requires downloading the points to a PC and correcting them based on satellite information received at nearby base stations, and that's with good satellite lock (perhaps 6 or more) and staying away from large vertical obstructions like trees and utility poles that deflect or block signals. Licensed surveyors can get accuracy at the centimeter level only by using very expensive antennae (USD$10k range) and occupying the same point for hours, taking an average of thousands of readings. And then as some posters here have said, vertical accuracy is still only about a third as good as horizontal. I think 10m with a phone is not impossible given that it is probably indoors or in a car, or while moving, and considering satellite lock comes and goes. The chip's primary duty is to supplement or correct info given to E911 dispatchers, and so it probably doesn't get much better than differentiating one single family home from another.

anchorite
  • 291
  • 3
  • 4
11

10CM accuracy is pretty unlikely. Military MGRS systems using the military GPS system, for example, which tend to be a bit better than civilian equivalents, have a 1m resolution. Assisted GPS may provide tighter resolution if available, but right now it takes a device significantly larger than a cell phone to achieve a 10cm resolution. The GPS resolution depends more on the hardware than the software.

phoebus
  • 14,673
  • 2
  • 33
  • 35
11

You can routinely get 2 meters accuracy with some Android phones, like the original Motorola Droid.

I have verified this in many ways over the past 1.5 years.

EDIT: And I sell a gps app with over 300k users.

Fraggle
  • 8,607
  • 7
  • 54
  • 86
  • 1
    Well first off there is the reported accuracy that you get with the Location object (locaiton.getAccuracy()). You can use apps like GPS Test (free version) to see what kind of results you get. I've tested this many times. I've also tested against known positions, and verified distances from targets with known positions. Anyway I could go on in some detail here, but not really the place for it. Needless to say, doing it perfectly can be tricky as anything you are comparing to may have errors in it as well. – Fraggle Dec 10 '11 at 18:39
  • I think it depends on being stationary or moving. It also depends how long you are stationary for. The more you stay put, the more accurate your location will get, right? – Radu Jul 26 '13 at 10:13
  • @Fraggle Maybe you can add links to articles which DO go into details/depth on accuracy in GPS co-ordinates. – GuruM Dec 16 '13 at 07:19
  • @GuruM don't know of any articles that do a good job on this. Lot's of players want to spin things their own way. Plus everyone's hardware and environment is different, GPS can even be effected by weather. On Android you can test yourself with "GPS Test" app. Or in iPhone the "GPS Status" app is ok. Always test outdoors clear of big buildings. Test on multiple days to get good averages, and give the phone a few minutes to get a satellite fix. – Fraggle Dec 16 '13 at 17:59
  • [Deep Dive into Location by Reto Meier](http://android-developers.blogspot.co.uk/2011/06/deep-dive-into-location.html) might be useful. [GIS stackexchange QnA](http://gis.stackexchange.com/questions/43617/what-is-the-maximum-theoretical-accuracy-of-gps/43657#43657) – GuruM Dec 17 '13 at 05:38
  • 3
    @Fraggle I think it's a good answer; what would be interesting is a list with phones and their performance on GPS accuracy; this should be very easily done by someone like you (because of the GPS app); – msysmilu Feb 23 '15 at 11:40