GPS, Running Watches, Phones, Accuracy and other stuff pt 2

Theories and first attempt at quantifying

In Part 1 I went over the background of this, the question is what is best to use for tracking runs (or any other activity) - phones, running watches or general fitness trackers.

I concluded that as the chipsets all promised similar accuracies,  then other factors may play a part.

My thinking is that there are a few things that could help or hinder

  • Physical size of device - the bigger the device, the bigger the aerial
  • The targeted design - Running Watches will, I hope, be designed to get the best signal, so may have bigger/better aerials than phones
  • The location/orientation - if you have your phone upright in a back pocket the aerials may well be orientated in the worse way for signal reception - a wrist watch should be better
  • Age of the device - runners who say "phones are rubbish" may have last tried using old phones, as lots of us have old phones to take out running rather than their newest ones.  Who wants to slip and break their new phone, when that old one is doing nothing.
So, here comes the nerdy stuff.  How to quantify accuracy?

Well, the obvious first case is does the tracker say the distance that you think it says - easy to do on known runs, like parkruns.  

I have a Samsung Gear Fit 2 - not a "running" watch, but a general fitness tracker.  It has GPS, but just the US GPS (AFAIK).  My parkrun distances it records are usually a bit under, but never more than 20 meters.

But looking at the plots, then you do see a different picture-

The red line is a highly accurate route, which is as good as the route could get.  The blue is this morning's trace (the parkrun has  two laps, the first is shorter)

You can see the issue - not great accuracy.  But overall the distance was fine.

Looking at the raw data, whilst there is a record per second, the XY only changes every 4-10 seconds



One working theory is that this infrequent updates actually smooths errors out.  

Looking at data I know was recorded by an old phone a few years ago - I don't get these identical points.


So, far so interesting, but surely with the tools I have, the internet, a work colleague who knows more stuff than I do and a quiet few days, I can work out something better to quantify this?  See Part 3. 

Popular posts from this blog

GPS, Running Watches, Phones, Accuracy and other stuff pt 5

So, I got myself a Garmin Forerunner 3 - a basic running watch from Garmin, good enough for what I need I also spent a bit of time re-plotting the route using the high accuracy OS Mastermap.  MasterMap in my neck of the woods is captured at 1:2,500 - which means it is 99% confident to be within 2.4 meters of the real world - see https://www.ordnancesurvey.co.uk/business-government/tools-support/mastermap-topography-support - without going out with survey gear and stuff, it is as good as you can get in most of the UK.  (Larger urban areas are captured at 1:1,250 but those be big scary places, like Exmouth) Using the new route I re-analysed the Gear Fit data and 4 runs with the Garmin, and here are the results - you can sort by the various measures (getting this working on Blogger took WAY longer than it should have) device error area samples device distance (Dd) OS distance (OSd) ...
Read more

Plex, Kodi and Logitech Media Server (LMS, Squeezeserver) on one Raspberry Pi 2

OK, mainly for my sake when I have to redo it. The issue.   I have a Home PC and a family The family use, in order of arrival. A Squeezebox 2 for music - the Squeezebox software and music being on the home PC Plex - for watching movies on other devices around the house and at Grannie's.  Plex Server is the Home PC Kodi -For watching movies and TV shows ripped from our DVD collection on the home TV.  Kodi is running on a Raspberry Pi 2 The Objective To have all these media services running on a Raspberry Pi 2.  True I know for Plex I'll have to stream the original  format of the movies rather than letting Plex re-encode depending on client.  This may change if I upgrade to a RP3 All have been setup elsewhere, but I want them all on one always on device.  Partly to let other family members use my Plex files without impactiung on the main Home PC and without me having to leave it on all the time. The Method I started with Plex,...
Read more

GPS, Running Watches, Phones, Accuracy and other stuff pt 3

Image
Enter FME FME is a massively powerful GIS data manipulation tool.  For professional use it isn't cheap, but those very nice people at Safe Software (who make FME) allow free Home use - if you ask them nicely and say what you plan to do with it.   I have a little experience with FME from work, and work with someone with a shed load more (many thanks Matt) so with my ideas and his pointers as to what tools to use, I have made this first attempt at quantifying accuracy. Let's look at a recent run The yellow line is a best fit line using open source mapping and google aerial images. The red is what my Gear Fit tracked. Now, my working theory is that by making polygons between these two (the grey areas) I should get a measure of how good the tracker was.  The smaller the overall area, the better the track was on that day The results for this first test are below - The overall Error Area (as i am calling it) is 56576.6 m2 - a lot of tha...
Read more