Hi there,
we included tracking in our project and mostly it works fine. In some cases, it jumps to one or several points that are located within the next town and back leading to are pattern that looks woven. Another tracking app that runs parallel did work normal, thus, it is not a problem with the GPS reception but within QField, seemingly.
Any ideas?
Thanks, Peter
@p.j.meffert , are you on a Samsung device by any chance?
One of the devices is a Samsung S20 5G, the other one a Sony Xperia 10. With my Samsung TAB 11 I had no problems, so far.
We are still struggeling with this issue. A collegue reports: “The location is displayed correctly, but when I tap ‘Start Tracking’, the point jumps to a town several kilometers to the north. Then it jumps back to my correct location and draws a kilometer-long line. If I delete the track and try again, it all starts over again.”
We are aware of this behavior when the Android mock location is used. Are you using an external GNSS antenna and the Android mock location?
We are using only internal antennas and no mock location (as far as I know). Just normal usage.
What about the GPS options “WLAN” and “cellphone net” (to improve the accuracy) – could these cause the trouble?
We tried, but it did not solve the problem :o(
Does turning on the accuracy restriction mode help with this? QField Settings > Position
You could set a bad accuracy threshold to 10 meters or something, so it allows for your ‘normal’ accuracy of the internal device but should not save vertices beyond that?
Or are these points not considered as bad accuracy by the device?
Hola, según entiendo estás utilizando una capa para grabar tu trayectoria. Yo tengo configurados mis proyectos para lo mismo y lo que me ha funcionado es en la configuración del seguimiento antes de iniciar es activar la opción “Salvaguarda de distancia errónea”
This behaviour happens a lot. The following settings help to reduce it, but it does not eliminate it.
When using another tracking apps besides QField these do not show this behaviour. We use TopoGPS as a backup. It would be a great releaf if QField can fix this.
Greetings. While I cannot replicate locally, it occurred to me what might be going on here is interference between satellite positions and other sources that might kick in (eg wifi scanning derived positioning)
I created a PR where we disable non-satellite sources for the internal GNSS device.
Could ppl in this thread who are impacted by this issue try the test APK and confirm whether the spikes are gone? APK download link here: Only rely on satellite positioning to avoid spikes when tracking positions by nirvn · Pull Request #7535 · opengisch/QField · GitHub
Thanks @Mathieu_Pellerin. We will test the APK this week.
@Mathieu_Pellerin Due to the bad weather my colleague has not yet been able to test the Pull Request in field conditions. So I have tested it in office conditions. I noticed that this feature was implemented in 4.2.2, thus I also tested the released version.
Firstly, I did not observe the spikes while tracking in both releases. Thanks for that fix. However, to my surprise, the Pull Request behaves differently from the official release. The Pull Request showed a better performance in tracking.
QField 4.2.2 seems to stop tracking when switching to another app for 30 seconds. This resulted in the track freezing or the tracked path was drawn as a straight line. Strangely, the recorded M-value reports s a time interval of 1 second .
The Pull Request actually did keep tracking. When switching back to QField it produced a neat tracking line with a vertex along the path and a M timestamp of almost every secon d.
For both versions I used the following settings:
Hopefully, you can find any difference in the code and find the source of this difference. As mentioned, I prefer the Pull Request, since field workers have to use other apps in the field to, for instance, identify species or check the weather forecast.
Screenshot Pull Request.
Screenshot QField 4.2.2:
@ckuipersatkb , hmm, I can’t think of a change that would lead to background tracking stopping after 30 second away. Can you reliably replicate this time after time?
@Mathieu_Pellerin I used 30 seconds to standardize the test.
Today I have tested it again by walking up and down in front of the office. As test I walked one way using QField and walked back using other apps. So the distance/time is the same.
For QField 4.2.2 I used another track than for the Pull Request. They are located next to each other along the same side of a canal.
The results:
I made screen recording to illustrate the behaviour. But can not upload them in this message. Is it possible to send them by mail?
@ckuipersatkb , yep send them over to mathieu@opengis.ch .
Could you do a similar test with the APK build attached to this comment please: Adjust tests · opengisch/QField@39e8a65 · GitHub ? Thanks
@ckuipersatkb , that is very interesting. The build I sent you is exactly the same as QField 4.2.2, except for the fact that it has a different app name (ch.opengis.qfield_dev).
That makes me think that the production version of QField you download from the play store has been flagged by your OS by some battery saving service or something along those lines.
Okay. Like you said interesting. Is there something I can help you with?
