-
-
Notifications
You must be signed in to change notification settings - Fork 362
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Notes may be positioned at places where the user did not intend them to be placed #5581
Comments
Can not reproduce (v57.2, but nothing changed in this respect, as far as I know). I long-pressed and created a note at the southern vertex of this house. The note pin dropped at exactly this position. I didn't move the map and the note pin stayed at this position regardless of whether I had the keyboard open or not. When the note was uploaded, it also appeared exactly where I tapped the map. https://www.openstreetmap.org/note/4199534#map=19/53.58427/9.94380&layers=N Maybe this is based on a misunderstanding? Were you expecting that the note is created at your current GPS position rather than at the position you tap on the map? |
This comment was marked as resolved.
This comment was marked as resolved.
This comment was marked as resolved.
This comment was marked as resolved.
I don't think so. The point is: I create hints/notes offline, then upload them to OSM, and eventually find them in https://www.openstreetmap.org/ with a significant offset. For example https://www.openstreetmap.org/note/4198670#map=19/49.19970/12.28992&layers=N |
Is this happening with all notes? If yes, you could take a screenshot of StreetComplete with the note visible before uploading and post a link to the uploaded note on osm.org to compare both. Another idea: is your map view in StreetComplete tilted? Or are you using the birds eye view (straight from the top)? |
I can also not reproduce it with a tilted map, nor a rotated map, nor when first placing the note and then while the form is open, rotating the screen. Closing as not reproducible. One thing I could imagine what happened is that due to the buildings being displayed in 3D, there was confusion on where the building actually starts. However, unlikely in an area with detached single family homes. What I find most likely at this point is that maybe the user's GPS location was off, and due to him not seeing the housenumbers (due to being zoomed out too much or the housenumbers just having been added and thus not being on the map yet), he assumed he was one building further down the street than he really was at that point. What people often don't know is that the accuracy circle drawn around the displayed position on mobile apps means that the there is (just) a 68% chance that the true location falls somewhere within this circle (with the center of the circle being the likeliest). So, the chance that the displayed location is actually outside of that circle is still about one third. (source) |
(Updated my previous comment) |
Yes, the current user position can be off more than displayed, not only due to GPS line-of-sight, but because majority of users have "Improve accuracy" enabled by default. And that option often does completely the opposite, easily throwing users location by dozens (and in some cases hundreds) of meters. So it would still be very interesting if @sjvudp could make a screenshot just before they click on that orange checkmark ✅ button to create the note, and then we can compare that to the actual position of the Note in the OSM map. Because, if they placed the note in the middle of some building, and it ends up on the other side of the street outside of a building, it is then the issue unrelated to GPS/location accuracy issue. |
I f have some interesting updates: As one of the examples given earlier also involved taking a photo, this might explain the issue if the position saved with the note is not the tap position, but the GPS position (which would be kind of nonsense IMHO), and the user finishes the note before the GPS position has returned to the original position. I also discovered another case for a note (Hinweis 4207189) created today: The Offset is about 25 meters! That makes me wonder: Doesn't Street Complete display the "circle of confusion" (HDOP, Horizontal Dilution Of Precision)? Even if not, couldn't it save it (or a corresponding value provided by the Android API) together with the note (so we could decide whether it's poor GPS reception)? |
Another thing I noticed was that Street Complete works with obsolete OSM data (for example see the parking area at https://www.openstreetmap.org/edit?note=4207189#map=19/49.19334/12.26630). |
Background map tiles show processed map data shown with delay, quests are based on live data (fetched via editing API, in case of conflicts edits are not applied)
what is saved is note position marked on map
Yes. You can also create notes instead answering a quest but then mapper cannot select location
this is based on GPS location as reported by phone |
Let's discuss that in separately opened #5590. Generally, for each separate problem, one should open separate GitHub issue (otherwise it becomes impossible to track what is resolved and what is not). If they are related, they can be referenced by putting e.g.
It does, if Android OS reports it. See for example this picture which shows it (but as noted above, that "circle of confusion" may be very imprecise)
If there was a need, one could create a separate version just for you which adds such debug. See for example #3312 for similar problem with notes location (also if you're inclined, you may want to read it to give you ideas how StreetComplete works / what one could try). We should collect some other information fist; see my questions at the bottom of this message
Ah, that sounds interesting. I'll try to play with that a little; there might be problems if you move while StreetComplete is not application in the foreground (e.g. while camera app is opened). What would be interesting is you @sjvudp could answer those:
|
I'll answer and comment some of the question:
That is very bad IMHO, specifically as OsmAnd works differently: In OsmAnd the position of the tap is the position of the hint.
For the case described I did not more (more than 50cm).
I did not enable "improve accuracy". |
note is placed on long tap location (or elsewhere if note was moved), or on object location if triggered from quest answer blue dot is based on GPS location as reported by phone
you can |
I think there is misunderstanding. I understood "this" in @matkoniecz comment "this is based on GPS location as reported by phone" to refer to "position indicator" (i.e. that little blue circle with white outline 🔵 showing where the phone thinks you are currently located at) and not to "note position marked on map" (i.e. that white exclamation mark in big red circle, which is a place where note would be placed). In other words: yes, you absolutely should be able to long at any location you wish and the map and leave an OSM Note there, regardless of where you are currently physically located (i.e. where little blue circle is positioned).
Thanks. I'd love to know if you can get some "background gpx track record" app working. If it happens like that, the problem likely lies with your Android phone system software (and should be happening with other foreground-location apps too). Tweaking power saving settings perhaps might help with that, but no guarantee (some instructions about quite complicated power savings might be found at https://dontkillmyapp.com/ for some more popular phone manufacturers) What would also help trying to reproduce the problem is if you can share "screen recording" video of the bug happening in real time @sjvudp . |
Heads-up: this might be genuine bug in Notes location (perhaps introduced around 57.x?); I've just experienced similar bug in SCEE at Helium314#543 where note was created some 200m off the location where I was creating it. |
Well, leaving aside whether there is a bug or not, but considering the interface is intended to be simple, the user interface probably needs improvements:
If the user tried to pan the map (maybe to see something on the map while entering text), the note will be re-positioned, most likely without the user being aware of it. (In contrast OsmAnd keeps the initial position, but allows to relocate the note in a later step. I think that would be much more preferrable.) If there were an undo function for notes, I would make some experiments, but I don't want to annoy other people with "test notes". I don't know whether a "GPS position walking off" has any influence on note placement, but I realized that switching apps causes exactly such "GPS position walking off". One example would be using StreetMeasure. For completeness I was using ColorOS 13.1 (Android 13, CPH2371_13.1.0.607(EX01) on a CPH2371_11 (Oppo Find X5 Lite) |
Hm well, you find the UI of OsmAnd better. From my point of view, it is just different. Maybe you find it better because you are used to it? Also note the text below "Create a note", it explains that you can adjust the note's position by moving the map.
You can turn off auto-sync in the settings. While the note has not been uploaded yet, you can undo it. |
we could delay uploading (the first 5?) notes and show an undo for 10 seconds. |
While familiarity of course always plays a factor, I too find OsmAnd way much safer. Even if I'm quite familiar with both, and I generally like StreetComplete UI better and easier to use, this is one of the exceptions: even when the user well knows that the note can be moved at all times, it is apparently too easy to move it unintentionally. To clarify, my metal procedure for creating the note works in separate discrete steps, e.g.:
The fact that the created note immediately disappears from the map in StreetComplete (while it remains visible at its final location in OsmAnd) compounds the problem, as the user does not see that the note has been created at the wrong position. In OsmAnd, if user happened to create the note at invalid position (which is much harder to do in the first place!), there would be immediate & permanent visual feedback, and user could correct. As for the solution that would fit into StreetComplete, I'm not sure. Perhaps:
It does... However I think it has been somewhat shown in the past that users won't always (to say the least) read such (especially multi-line) explanation texts 🤷♂️ - but that is not the main problem here, I think (well, at least it is not in my case). |
Another example for a severely mis-place note from this afternoon, making the note practically useless: https://www.openstreetmap.org/edit?note=4224996#map=19/49.18939/12.27518 |
can you capture video of note creation process? or is it happening only very occasionally? |
I found this related issue, where pressing the compass icon (maybe even accidentally) messes up note location: #3101 |
The point is: It does not occur all the time, and videos fill the storage quite fast (especially when long or many). |
Well, if given video has not captured it then you can delete it rather than keeping it forever. So storage is needed for one screen capture. |
To make things easier, it would be nice if a note would be displayed a few seconds (like 3) after it was filed, so the use could see where the note was actually placed, and if he/she could tap on that marker to re-open or re-position it, it would be nice. |
Just for info - I uploaded a bunch of notes with pictures a couple of days ago, from https://www.openstreetmap.org/note/4252536 onwards. In most cases notes were actually uploaded later, due to lack of phone signal. You can see that the note numbering is sequential even though the notes are hundreds of meters and several minutes apart. In all cases the note position was correct - I recorded Garmin waypoints at the same time, and the positions match within a few meters. I suspect that the reported problem here is more likely due to the user's phone misreporting the actual position (which can happen for all sorts of reasons). Perhaps the reported could verify that with some (external to StreetComplete) position tracking software? |
@SomeoneElseOSM It's hard to tell whether the phone mis-reports the current position. I used several apps to verify the position accuracy, and it looked all very well. I don't know how the API to read the current position works, but I could imagine the API being used incorrectly, maybe. Specifically if there are multiple "consumers" (i.e. different apps requesting the position at different intervals). |
How to Reproduce
Walk along, place an OSM hint/note.
After upload to OSM, the hints are significantly misplaced, For example: #4191986, #4191985
Maybe the reason is that position updates are seemingly "damped" a lot
Expected Behavior
Hint will be filed with correct "long-tap position"
Versions affected
v57.1
The text was updated successfully, but these errors were encountered: