Inaccurate speed and distance (records high for both)

[GoogleCodeExporter]

What steps will reproduce the problem?
1. Go on a walk (low speed usage)
2. Record a track
3.

What is the expected output? What do you see instead?
The speed and distance do not match up with my handheld garmin (I think the
Garmin is very accurate).  They are significantly inflated (approximately
20%). When I look at the map of my track, the track shows up as a zig-zag
line.  I downloaded endomondo and it seems to do a much better job.  I used
them simultaneously today (I'm on a trip and don't have my hand held
garmin) and Mytracks recorded 2.06 miles while endomondo recorded 1.69.  So
Mytracks records 22% higher.  Mytracks records my walking speed as 5.12 mph
overall and 5.68 mph moving average.  There is no way I walk that fast. 
BTW, Endomondo asks what activity I'm doing.  I don't know if it applies
different filters based on that activity.

What version of MyTracks are you using? On what version of Android? On what
phone?  Mytracks version = 1.0.15
Android version 1.5  (not much longer!!)
phone = Motorola CliQ

Please provide any additional information here:

Original issue reported on code.google.com by wru...@gmail.com on 5 Jun 2010 at 3:13

[GoogleCodeExporter]

I am getting the same thing with
HTC Hero (running android 2.1)
My Tracks version 1.0.15

Distances and speeds seem abnormally high, rendering this tool useless :/

Original comment by Aaronofk...@gmail.com on 5 Jun 2010 at 5:09

[GoogleCodeExporter]

These are the steps I would suggest for fixing this:
1. Create a new preference for the smoothing factor.  Look in 
res/xml/preferences.xml.  I would base it on the 
IntegerPreferences where the choices are 1 {off} up to about 5.  With a name 
something like 
LocationSmoothingFactor. If the value is 1 basically use the old behavior.
2. Use that factor as the size of the smoothing buffer used in the 
TripStatisticsBuilder.  
{http://code.google.com/p/mytracks/source/browse/MyTracks/src/com/google/android
/apps/mytracks/stat
s/TripStatisticsBuilder.java} You will see similar examples for the speed and 
elevation.  I think we would want 
two new DoubleBuffers.  One for lat, one for long.
3. We calculating distance in the addLocation method.  Instead of calculating 
the distance from last location to 
current create a synthetic lastLocation, then add the data and create a new 
syntheticLocation.  Calculate the 
distance between these two points.

Original comment by sandordo...@google.com on 6 Jun 2010 at 5:46

[GoogleCodeExporter]

When I send my track to Google Maps and hovering over the route in Google Maps 
shows the much smaller distance that I believe to be accurate.

Original comment by mbhaff...@gmail.com on 18 Jul 2010 at 12:52

[GoogleCodeExporter]

Thank you mbhaffner for your nice verification of the problem. 

Question to anyone:  Do these issues get looked at by the programmers? 

Original comment by wru...@gmail.com on 20 Jul 2010 at 9:49

[GoogleCodeExporter]

wrusry > The developpers are listed here: 
http://code.google.com/p/mytracks/people/list. One of them participated to the 
present discussion, so your answer would be yes.

sandordornbush > Thank you for this app. It is great and I like it even more 
now that I know that it is open-source.

I have noticed the problem mentioned above when hiking on marked trails. 
MyTracks would systematically record larger distances than the advertised trail 
length.

Original comment by florent....@gmail.com on 3 Aug 2010 at 5:22

[GoogleCodeExporter]

Specific example of issue:
Go on a bike ride.
Total Distance: 80.64km (before it stopped recording by itself)
Total Time 29:02, Average Speed 166.62km/h
Moving Time 3:42:46, Average Moving Speed 21.72km/h

1.0.19, Android 2.2, Google Nexus One

Original comment by matt...@gmail.com on 17 Oct 2010 at 8:52

[GoogleCodeExporter]

Original comment by sandordo...@google.com on 17 Nov 2010 at 2:38

• Added labels: Targeting-1-1

[GoogleCodeExporter]

I have yet to test this change in the wild but I think the fix is in this 
revision.  Please review at:
http://code.google.com/r/sandordornbush-bugfix-mac/source/detail?r=99a5ac82e105f
eaedbfe6c245b2eeedc8fc01371

Original comment by sandordo...@google.com on 6 Jan 2011 at 6:59

[GoogleCodeExporter]

I would be glad to test this.  Is there any way I can install this on my phone 
prior to a released version which contains this change?

Original comment by wru...@gmail.com on 6 Jan 2011 at 2:29

[GoogleCodeExporter]

I realized this AM that this is not the right fix.  I am going to revert this 
and do it slightly differently.

Original comment by sandordo...@google.com on 6 Jan 2011 at 4:43

[GoogleCodeExporter]

Thank you for your efforts!

Original comment by wru...@gmail.com on 6 Jan 2011 at 4:51

[GoogleCodeExporter]

I pushed a new version of this change and I have tested it a little bit.

Rodrigo can you check it out.

Original comment by sandordo...@google.com on 13 Jan 2011 at 5:45

• Added labels: Work-Small

[GoogleCodeExporter]

This will be in the next release.  We will send out a release candidate soon 
for people to play with.

Original comment by sandordo...@google.com on 14 Jan 2011 at 9:36

• Changed state: Fixed

[GoogleCodeExporter]

That's excellent news. Thank you!

Original comment by florent....@gmail.com on 14 Jan 2011 at 11:36

[GoogleCodeExporter]

I'm wondering how was this issue resolved. Issue 340 reports the problem as 
still existing http://code.google.com/p/mytracks/issues/detail?id=340

Original comment by kenep...@gmail.com on 5 Jun 2011 at 8:33

[GoogleCodeExporter]

I went on a hike recently and used MyTracks version 1.1.5. As I reached an 
indicator that stated that I had been walking for 5 km, MyTracks was indicating 
7km. That seems like a big discrepancy!

I haven't yet had a chance to see if MyTracks 1.1.6 still has the bug.

Original comment by florent....@gmail.com on 5 Jun 2011 at 8:49

[GoogleCodeExporter]

I just went on a urban walk and recorded 2.52 miles with MyTracks and 2.21 
miles with endomondo on the same phone at the same time. Endomondo had 
previously performed the same as my handheld garmin GPS.  

This is with Mytracks 1.1.6.  These results are about 14% different.  That's 
lower than when I started this thread, but I think still not where it 
should/can be.  I will continue to test it.

Original comment by wru...@gmail.com on 6 Jun 2011 at 11:37

[GoogleCodeExporter]

This inaccuracy is still occuring with MyTracks 1.1.6. Urban walk over 7.6 km, 
but My Tracks reported 9.2 km - approx 21% high. Similarly high with its 
average speed reports every km.

This is on a HTC Design HD, Android 2.3.3 (also occurred prior to 2.3.3 
upgrade).

Original comment by womb...@gmail.com on 13 Jun 2011 at 7:21

[GoogleCodeExporter]

My continued testing also shows about a 20% too high distance/speed.

This seems like it deserves to receive high priority in the bug list.  Or maybe 
it is a feature and people like being told they are running/walking faster than 
they really are.  :)

Original comment by wru...@gmail.com on 13 Jun 2011 at 7:28

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

There is a consensus this issue is outstanding. At number 77 it was an early 
issue and I'd agree with comment 19 this issue deserves high priority. 

Accuracy depends on how total distance is defined. I discuss this at 
http://groups.google.com/group/mytracks/browse_thread/thread/3db73dca27510742/c5
8a7e8b92b9f7e9?lnk=gst&q=author%3A%28Ken+Taylor%29#c58a7e8b92b9f7e9 and propose 
a definition. Comment 16 has a sign showing 5km when My Tracks reads 7 km but 
in the bush distance is hard to measure so how was the sign placed. Was it 
straight line distance from the start or was it estimated from a map ignoring 
the twists and turns in the path i.e. what did they mean by 5km?. Comment 17 
equates accuracy to the same as Endomondo which is equated to a Garmin GPS. 
Consistency between applications is important but if there is a difference the 
one which is the more accurate depends on the definition of distance adopted. 
I've read elsewhere there is differences between different models of Garmin 
GPS. Comment 18 reported an urban walk distance and compared My Tracks to that 
distance getting a difference of 21% but what definition of distance was used 
for the urban walk distance and how was it measured. I tried using the Google 
maps "Get directions" distance which I assume is the centre line of the road as 
the reference distance in the discussion at 
http://maps.google.com.au/maps/ms?hl=en&ie=UTF8&oe=UTF8&msa=0&start=0&num=200&sl
l=-35.663514,149.161339&sspn=0.079908,0.005868&t=h&z=11&msid=2172542005655117041
18.000484626f3019359f03a . This meant I implicitly decided to exclude wandering 
around at rest stops like this one 
http://maps.google.com.au/maps/ms?hl=en&ie=UTF8&oe=UTF8&msa=0&start=0&num=200&sl
l=-35.663514,149.161339&sspn=0.079908,0.005868&t=h&msid=217254200565511704118.00
0484626f3019359f03a&ll=-35.614783,149.152464&spn=0.000829,0.002087&z=19 . Using 
this definition of total distance I got an overestimate of 12% prior to the fix 
in comment 13. Comment 19 says distance is 20% too high but too high according 
to what definition of distance? Also whatever the definition chosen, the 
percentage error will be different for handsets with different GPS chip sets of 
varying accuracy, though GPS sample error should be consistent across 
applications. I tried comparing total distance between different applications 
at 
http://groups.google.com/group/mytracks/browse_thread/thread/8ee8da5686700191 
and the My Tracks total distance was greater than both the others. I also 
noticed that the Google My Maps algorithm had some desirable properties, the 
most obvious being that total distance was robust to the different GPS sampling 
strategies employed by the different applications. My Tracks gave a total 
distance 5% greater than My Maps.

Original comment by kenep...@gmail.com on 20 Jun 2011 at 5:23

[GoogleCodeExporter]

I think that the problem is in inaccurancy of the GPS receiver in phones 
(hills, forest, tall buildings). I'm walking straight beside a river, and the 
recorded track is not straight. The application records about 20% more in speed 
and distance.
The recording algorithm requires some GPS data filtering 
(averaging/smoothing/etc) - at user configurable level. There was such option 
on my Nokia phones track recording application (Nokia Sports Tracker).

Original comment by vadg...@gmail.com on 21 Jun 2011 at 12:39

[GoogleCodeExporter]

I know this is a quite old thread, but i think the problem is not solved yet.

What steps will reproduce the problem?
1. Go running (low speed usage)
2. Record a track
3.

What is the expected output? What do you see instead?
When i go running i normally go more or less straight. MyTracks tends to record 
tracks with zig-zag line elements. 
Once this resulted in 19km recorded track for about 7km real distance. I don't 
have any hills or tall buildings close to the track. GPS precision to be 
recorded was set to 200m (default).

What version of MyTracks are you using? On what version of Android? On what
phone?  
Mytracks version = 1.1.12
Android version 2.3.3
phone = HTC Desire Z

Please provide any additional information here:
http://maps.google.com/maps/ms?msid=211297825871202552287.0004b4217d8aef7ac31d7&
msa=0&ll=41.401934,2.213847&spn=0.002611,0.005413
shows the phenomenon at Moll de Mestral on the way back, but not on the first 
passage of the very same spot.

Original comment by oli.drec...@googlemail.com on 25 Dec 2011 at 10:03

[GoogleCodeExporter]

Yes, I do not know why this bug does not get any attention.  I use another 
program because of the inaccuracy of MyTracks even though Mytracks has better 
features. Perhaps people like to think they are going faster than they are.  :^)

Original comment by wru...@gmail.com on 25 Dec 2011 at 5:48

[GoogleCodeExporter]

April 2012 and the same problem still present. Why not just close up shop and 
pull this thing from the market instead of giving people a crappy experience 
under the Google brand? You're just diluting the brand. 

Original comment by amaury.rodriguez on 14 Apr 2012 at 10:51

[GoogleCodeExporter]

This issue is still live.

Evo 3D

Original comment by pul...@gmail.com on 17 Feb 2013 at 10:59

[GoogleCodeExporter]

Same here, why not use multiple GPS samples and create a mean out of them and 
use that as position? Really no improvement since years :-(

Original comment by m...@christoph-petersen.de on 10 Jul 2013 at 4:21

[GoogleCodeExporter]

One way to avoid the zig-zags is to first get a GPS lock using the cell towers, 
then go to airplane mode, and then restart the my tracks app. This has never 
failed. However this is a really bad solution if you want to be able to receive 
calls while recording the travel. The obvious solution that the Google 
programmers could use is for My Tracks to use the cell tower info to get an 
initial GPS lock, and then only use the satellite info after that. The Garmin 
uses only satellite info and never does a zig-zag. 

Original comment by slowrunn...@gmail.com on 10 Mar 2014 at 9:48