Beta 14 release of the Speed Reader tool is up at www.ka72.com/

This version provides the first cut of GPX file support. I would appreciate some more test data to work with. I only have a couple of files. Any tips on identifying bad files would be good too (for instance, how to spot a Garmin in "Auto" mode.)

This version also improves data filtering, and adds a 100m speed to the results (by request.) The Google Earth overlay is improved, though I have to purge all the old overlays to force them to be recreated, so if you have been using a link direct to an overlay page (like the one I posted at gpsteamchallenge the other day) it may stop working until you revisit the site and reprocess the file.

I'm almost ready to start working on the Speed Locker site, just a couple more engine things to do, but I wanted to say a massive thankyou to everyone who has helped out by trying out the core SpeedReader tool, and I wanted to assure everyone that the tool will remain available and free for the forseeable future (though I will probably drop a PayPal donate button on the page if hosting costs go up.)

Dylan.

That's fantastic Dylan

I got a 2 sec peak of 52 knots on one of the files, and a 100m speed of a couple of hundred knots so some work still needed... but it's definitely getting a lot closer


Well done.

To spot a garmin in "Auto" mode, if more than half of the points aren't 2 seconds apart, then you've got a garmin in "Auto" mode. If more than half the points are 2 seconds apart then it's in the proper "2 second" mode.

Select to expand quote

nebbian said...

That's fantastic Dylan

I got a 2 sec peak of 52 knots on one of the files, and a 100m speed of a couple of hundred knots so some work still needed... but it's definitely getting a lot closer


Well done.

Great speeds there Nebs

Slowie will be quaking in his booties

Select to expand quote

nebbian said...

That's fantastic Dylan

I got a 2 sec peak of 52 knots on one of the files, and a 100m speed of a couple of hundred knots so some work still needed... but it's definitely getting a lot closer

Yes, I just don't have enough test data. There are disclaimers all over the place about not relying on the GPX results yet

Those Garmins seem to record massive speeds between tracks, which gives some interesting spikes. I've been playing with a way of ignoring those sorts of data items (like GPSResults does) but haven't finessed it into working right yet.

I thought it was good to get it up there and show that it is started, even if it is not reliable yet. Plus there was some other good stuff in the release.

Here's an interesting one for you Dylan,
Serial Number: 1000374
User Name: MIKE
Device: Navilink.GT11
V1.62(B1115)
Session Date: 4/02/2009
Number of Points: 23447
Approx Timezone: 7.5
2 Second Peak (kts): 317.38
5x10 Average (kts): 68.49
Top 5 5x10 speeds: 197.8406
44.0621
34.0231
33.2913
33.2543
1 Hr (kts): 21.41
Alpha 500 (trackpoints): 37.9
Nautical Mile (kts): 31.81
100m peak (kts): 344.96
Total Dist (dop km): 233.37

While I was having a break, I left my GPS in my helmet, which I had strapped to my boom. When I put the GPS in it had 55NM and max speed of 30kts. When I removed it had 94nm and 362kts. I've removed the mysterious excursion from my realspeed file, to get valid results, but wondered how well your site would cope with it. May be a hard one as it seems to have affected doppler data.

233 KNOTS!

I saw that one, and another with a similar bizarre high reading. Neither of them could be removed by my software.

I'm definitely not in the business of creating a GPS file editor, rather my plan is to make these sorts of files available for validation by competition officials. There are always going to be instances where hardware errors mean that you really need a pair of human eyeballs on the data.

Dylan.

Yes, agreed, you just need a flag to come up when stupid numbers appear.

Dylan,
I ran a GPX file this morning as it had a spike within the NM and wanted to compare with Realspeed results.

Everything was spot on within a decimal point except the 5x10 which was way out, 34.019 as opposed to 46.73.

Our captain ran the file and got a different NM to me as well so need to talk more to him as well, we are both running Realspeed, latest version.
He may have cleaned out the spike, I couldn't & as it was a zero I didn't think it would matter anyway.

Select to expand quote

Paul Kelf said...

Everything was spot on within a decimal point except the 5x10 which was way out, 34.019 as opposed to 46.73.

Thanks for the feedback.

I have made some terrific improvements in the gpx file analysis in my current development version, but I've also been working on the site infrastructure, so I can't send the update online just yet. 5x10 was one of the things that I needed to fix, and also Alphas were very inconsistent too.

The updated engine will go online in the next few days, depending on how busy I am, and a post will be placed here at that time. GPX remains experimental, but it is improving all the time.

Dylan.