Minimize GPS-errors on Tracking

franc · August 08, 2012, 22:46:51

Hello

it happend some times to me, that I tried to track some hikes with Locus and my GPS module ran mad. I had some jumps to the ivory coast in africa which made the track worthless because of total km and speed etc.
I wonder why this cannot be handled by Locus with some calculations. E.g. the gps location can found with mobile station as well, so when suddenly I might go some 10.000 km in some seconds (going on a special rocket probably

Locus tracking could double check with the position of mobile data provider, which is not exact but much better than those big gps-errors of the module.

Couldn't it be developed, that I have a setting, e.g.: "Don't allow unrealistic GPS-positions"?
Maybe with even some parameters like max. realistic speed, max realistic distance, etc.

I think this would help to make the tracking of Locus much better and more special.
At the moment I don't use Locus tracking because it uses nearly double the power than MyTracks, but if the tracking were much better, I would accept it.

Thank you.

frank

Menion · August 08, 2012, 23:28:12

hmm weird, there already checking for such nonsenses. I'll check it once more tomorrow. is this happening to you often? I'm sure you know, that you can simply delete such wrong points after recording anyway I of course agree that this should not happen in any cases

franc · August 08, 2012, 23:43:26

Sometimes this happens, I have to reboot my phone then, after it won't happen for a while. I think this is a combination of OS and GPS-module of my sgs2, but I don't know.

I never tried to edit tracks in Locus, as I normally use MyTracks to track. I measured both in battery consumption and detected that Locus tracking is nearly the double. Is this wellknown or maybe an error in my measurement?
But when my gps-module starts to run mad it is not much sense to edit anything, because I won't get reasonable values after it once start to fly to africa

Menion · August 09, 2012, 20:41:29

I'm checking code in Locus and I already have there protection against these faults. There is check that test speed needed to move from last to current location (so just distance-diff / time-diff). Testing value is 3000 m/s, anyway errors you talk about are much bigger right?

Would you be so kind and record for me NMEA message with such error from one of your rides? In NMEA message (enable this in GPS settings, file will be stored in data/NMEA) will be all raw data from GPS so I can simulate exactly whole your ride. Thanks!!

tommi · August 09, 2012, 20:49:55

Quote from: "menion"There is check that test speed needed to move from last to current location (so just distance-diff / time-diff). Testing value is 3000 m/s,

Hi Menion, 3000m/s = 10800km/h. Did I miss an invention? Or is Locus elected to be part of the instruments for the next mars mission :lol: ?
Btw. I tried GPS in a flying airplane, though some satellites were found, there was no sync. Do you know the speed limits? Is it a restriction of the chip or are there limitations in the algorithms in the used application software?

Menion · August 09, 2012, 21:08:37

hmm I was just thinking that no-one will move faster then 10800km/h with locus and track recording, or am I wrong?

Anyway when GPS send data every sec, so then I expect (and also as I saw on some screen), these errors are really huge and not just below 3km/s ...

btw. I've had never fly, so don't know

I hoped that I'll test it soon also, but you say it not work in airplane? damn ...

franc · August 09, 2012, 22:22:10

Did you get my nmea file? I sent it per mail ("track to africa"). I thought I posted it here, but don't find the post now.

I flew once with activated GPS module and tracking (the great App "GPSLogger" on a BlackBerry 8900) beside the fact that the module had an maximum altitude which was much fewer than the real one, the tracking was sucessful. Speed and location was correct.
I was near the window and had the phone hidden in my pillow, which I pressed to the window, as it is strictly forbidden to use smartphones, any phones in the plane nowadays. Only laptops with sim-card and HSPA, BT, WiFi etc

Menion · August 10, 2012, 08:44:53

thanks, NMEA received ...

your GPS has definitely some problems. Correct NMEA message line with coordinates should looks like

Code Select

$GPGGA,123519,4807.038,N,01131.000,E,1,08,0.9,545.4,M,46.9,M,,*47
since begin, you have in this NMEA such values, but in the middle, log is full of nonsenses, where instead of 4807.038, you have 4807,038 so data are useless because "," is used for separating values

I'll anyway try to check usable data that are included. Next report will come soon ...

franc · August 10, 2012, 08:58:47

Thank you!

Quote from: "menion"...your GPS has definitely some problems. ...

Time for a new phone

Menion · August 10, 2012, 09:47:41

this will be probably problem of software, not directly GPS device. If you use some custom ROM, suggest to change it

---

after some testing: damn, there is so many weird jumps. Just check from line 32486. Suddenly there is 0 latitude, after a while 12, after a while 13 ... wtf? ... somewhere in the middle I see again 25

PS: and just my personal note: I still don't had in hands better phone then SGS2

PeterPablo · August 10, 2012, 12:36:34

I experienced the very same issues serveral times, when using Locus Pro but never came around to report the bug. I have a SII with original ROM 4.0.3. I can not send you such a bad track, since I deleted all of them. Post-processing was no option for me.

The fastest airplane for public transportation used to be the Concorde (http://en.wikipedia.org/wiki/Concorde). According to wikipedia the top-speed was around 2200 km/h. So you should be safe by setting the threshold to 2500 km/h.

Hopefully you suceed in adapting your NMEA-parser to also cope with different seperators and filter out some nonesense values like extreme jumps in lat / long (in addition to your distance-check).

Thank you!

PeterPablo · August 13, 2012, 11:25:34

Yesterday I observed again of the above described glitch. One point jumped a couple of hundred km south-south-west of my position. This happened while I had bad GPS reception. In case you are interested in the recorded track-file I will gladly send it to you via email.

Peter

Menion · August 13, 2012, 11:47:48

problem for me, is that I wanted NMEA record from franc, but seems that his device also store incorrect NMEA messages so it's not possible to use it for testing. You can send me you file, or you may check for me what values are at this weird point compare to previous and next recorded points. Like speed, altitude, etc. Or I can check it by myself if you may export it to GPX file for me, thanks!

PeterPablo · August 13, 2012, 13:50:01

I sent you the exported track via email. I had NMEA messages disabled but switched them on now.

Menion · August 13, 2012, 13:52:34

Peter, thanks for track!

hmm seems that these values are much long-time ... I think, that it's not one location with wrong location and next second good location again (as I expected), but ...

here is example from your track

1. location 17:27:54
2. location after 52seconds! in 17:28:46, distance to previous 125km (so it's 2400m/s, or 8650kmph)
3. location after another 40s! in distance 119km (2975m/s)

this third location is anyway in distance 6km to first location (time diff is 92s, it's 230kmph) so it's almost real

I'll then try to create better checking - "double check"

firstly compare current location to last valid (treshold will be reduced to some more real value - like you said 2500km/h) and then also check new location to previous invalid location. We'll see if this helps anyway I'm little worried that still some weird location will pass through