Android GPS Accuracy Test ( Is that a 9i or a PW? and can I really trust this thing? )

[CR McDivot]

Do you really know how accurate the GPS is on your android device? Would you like to get some real world stats?

I found an android app, GPS Benchmark , that is easy to use and produces a plethora of information presented in both spreadsheet (csv) and visual (Google Earth) format.

I ran a series of tests on an older Cliq2 and got a lot better idea of just how she performs.

The app identifies the test location as “Ground Truth” the coordinates of which may be entered manually, by touching a point on a satellite image from the app (requires data connection), or even from a QR Code image.

I chose the latter, as I wanted to run a series from different locations and my fat fingers don’t get along too well with virtual keyboards let alone tiny sat images.

On my home computer, easily identified testing points were selected on Google Earth and the coordinates recorded. I used QRdroid Code generator to produce 5 geoQR s which I printed.

The GPS on the android device was allowed to aquire sats before being taken to the test site. Once at each “Ground Truth Location” the GPS Benchmark app was started and the test allowed to run for 1 to 2 minutes. MCT locations are from a separate Magellan handheld GPS.

The kmz file is best downloaded and opened in GE to see all the data from the tests.

( link to GPS Test.kmz )

[cristphoto]

I have a Bushnell laser and a Bushnell gps watch. A friend has the same watch. When the pin is near the green's center all three units are very close in their readings. When we play with someone that uses a cell phone often their readings differ (sometimes by a good amount). I wonder if the phones use different satellites to get their readings.

[TD22057]

Here's my understanding:  most (if not all) phones use a fairly low precision GPS chip.  To get good GPS data, you one that get read wide area augmentation signals (good article here: http://en.wikipedia.org/wiki/Wide_Area_Augmentation_System).  SkyCaddie does this as well as read Japanese and European equivalents ( https://skygolf.com/global%20repository/support/faqs/Can%20the%20SkyCaddie%20receive%20signals%20in%20countries%20outside%20of%20the%20United%20States.aspx).

However - phones also use cell tower locations to add information to their GPS solutions.  So some times a phone is going to have better accuracy than GPS+WAAS - but then some other times it won't.  It would depend on the number of towers and the geometry to them so it could change dramatically over a single course.

[CR McDivot]

Here's my understanding:  most (if not all) phones use a fairly low precision GPS chip.

The GPS Benchmark app tests the inherent accuracy of the chip, hardware, and OS. If any augmentation is used it will be noted.

The tests I ran were purposely without wifi or cell tower augmentation.

Played a round today with an app I typically use only for scoring. This time I compared distances and shot tracking with a laser, and came up with results generally in line with the expected 95% assurance. Some few readings considerably worse than that.

Having done mapping work for the state highway dept, using top end differentially corrected gear, I am well aware of the limitations of even the best systems and methods. I am surprised at the accuracy potential of the smartphone, but trust my laser infinitely more than even a dedicated handheld.

[TD22057]

The GPS Benchmark app tests the inherent accuracy of the chip, hardware, and OS. If any augmentation is used it will be noted.

Cool - I didn't realize the app would let you select the measurement sources to use.  I'll have to check that out.  I wonder how accurate the absolute Google maps coordinates are?  I found a few papers that did measurements in Africa and the Middle East and found ~1.5-3 m in Google maps but that may not be representative.

[CR McDivot]

...I found a few papers that did measurements in Africa and the Middle East and found ~1.5-3 m in Google maps but that may not be representative.

On the day, and at the location of my original test, Google Earth registered within 1m of a point collected by a quality handheld with a powered external antenna averaged over 3000 readings. This seems to change over time as historical aerials are not as well registered. Currently the best publicly available aerials are claiming 1 foot (primarily in US urban areas).