Gps sample rate and noise adding to wrong distance and times.




I posted something in feedback before seeing this but I figured I would past it in here too. 

""Hi, Ive spent a lot of time trying gps logging apps this week, and though I want to use Strava, I cant get past how noisy the gps data comes out compared to Runkeeper, Nike+ and even my old Garmin Quest.   For the most part I think it is great that Strava samples as fast as they do for especially fast activities, but because of the limitation of phone gps consistency, it is all over the place and adds the extra vectors for a total of .1 to .2 miles per mile. So while my garmin, runkeeker, and Nike+ all say the distance was 2.3 miles, Strava is saying it is 2.5 or more just because of the added noise.   This could be easily fixed by just collecting 3 or 4 data points and averaging some of the data I would imagine, or perhaps even just lower the sampling rate?  Would it possible to add an adjustable setting in the Android app to affect this?  I tried the Strava Snap site but because some of the area I was traveling didnt have a trail to snap to, it does not work.  I attached some pictures of the other apps reckoning of position data that was recorded at the same time with Strava(same walk) to compare vector lengths and over all smoothness. Nike and Runkeeper were on the same phone simultaneously with Strava while the Garmin Quest was on its own.  The first Pic is Nike+(green)over Strava(purple). The second pic is Runkeeper over Strava. The third pic is Strava(purple) over the Garmin track. Notice how short the vectors are for the Strava output.Notice in the first pic how much longer the vectors are than the Strava vectors in the 3rd pic(almost 4 to 1). Also, the Nike+ and Runkeeper were almost identical in output. Perhaps and averaging or longer sampling interval would reduce some static and make the distance accurate? Thanks

4

Commentaires

6 commentaires
  • Wow, I was just looking at Strava's heat map, and the slide feature. This would be an amazing way to simply correct GPS data.  Perhaps Strava can use the data you upload for a run, and rather than replace the data with the slide edited path, merely leave the original data to be incorporated into the heat map, and show the user in the activities page, an adjusted version that was corrected with slide, which has the corrected distance and timings etc.   This is better than the user simply adjusting their gps tracks to match the OSM map, because the OSM map may not be accurate, and itself needs to be adjusted to the Strava heat map  eventually.  This would be a great addition to the online editor? :)

    0
    Actions pour les commentaires Permalien
  • Runkeeper seems much closer to Strava than the others.  It appears that some of the twists and turns reflect reality that Nike and Quest skip over.  Runkeeper agrees that there are sharp turns (switchbacks?) at the top of the picture.

    A better test would be to borrow an 920XT and compare.

    0
    Actions pour les commentaires Permalien
  • Well, the reason Runkeeper agrees the most with Strava in that location is because they are poling the GPS with similar frequency, and the data in that area is a bit flawed from perhaps tree cover. The point isnt to compare a quality mapping GPS vs the recreational GPS on a phone, but to look at the differences between Strava, Runkeeper, Nike, and other mapping applications when poling from the same GPS at the same time. If you compare Strava to Runkeeper as you look around the whole track you will notice that for most of it Runkeeper is a bit smoother than Strava, yet not as smooth as Nike.   The real issue arises when the staggering and jumping around is so drastic that your distance is miscalculated as is the case for Strava. Runkeeper is orders of magnitude better than Strava, and the final distance is fairly accurate with reality. If Strava had a good way to repair the track after the fact, It would not be as big a deal. If they eventually incorporate the ability to use the heat maps to snap your route, this will be a huge advantage.

    0
    Actions pour les commentaires Permalien
  • Since this problem also bothered me I wrote a small application that simply takes a running average over 9 datapoints. You can have a look at https://stravasmooth.herokuapp.com or https://github.com/jonderwaater/stravasmooth. This solution is not ideal but it works for me; at least until Strava improves the situation.

     

    -1
    Actions pour les commentaires Permalien
  • If you're on a trail with a lot of sharp curves, switchbacks, and hairpin turns (e.g. mountain biking), whether you're running, biking, or hiking them, then you want the most frequent GPS sampling intervals possible.  Otherwise, and since these applications calculate distance as the straight-line vector between two GPS sampling points, your mileage will be calculated incorrectly and less than actual mileage as the distances around curves and turns are cut off.  All methods (ceteris paribus) will all produce similar results over a relatively straight course or one with long, swooping curves.  But just because two Apps produce similar results does not make them right.  The best way to measure any course is with a mechanical odometer like a surveyor's wheel, (a.k.a click-wheel, hodometer, waywiser, trundle wheel, measuring wheel or perambulator).  Since many bikers have odometers with mechanical sensors on their bikes, you might want to check with local bike stores and clubs to get their measurements on all kinds of courses (not just bike courses), then compare your Apps' measures against the bike's mechanical measures.  Once you see the difference between distances over "turny" courses as measured by mechanical wheels Vs Apps then you'll see you want shorter GPS sampling intervals, not longer / more samples, not fewer.

    0
    Actions pour les commentaires Permalien
  • You can also see how GPS sampling intervals cut off the distances around curves by using your GPS App to measure the distance of a winding course while on a bike, moving fast, and comparing the distance as measured while walking the same course.  While walking, the App will have more time to see and capture / sample your actual trek around the curves. I'm older now, with a lot of compromised bones, mostly hike the same trails I used to bike, and you might be surprised how much more distance the GPS Apps add when walking, giving them time to more accurately see your actual route.

    0
    Actions pour les commentaires Permalien

Vous devez vous connecter pour laisser un commentaire.

Ce n’est pas ce que vous cherchez ?

Nouvelle publication