Android Auto: movement direction map orientation is flipped and jiggles

[ZeLonewolf]

Description

When using Android Auto, driving south in map direction mode, the map is upside down and rocks back and forth.

Your Environment

Samsung Galaxy S22+ 2022 Chevrolet Spark Android Auto Android v13

Live video while sitting at a stop sign

https://user-images.githubusercontent.com/3254090/209214066-36741312-e9fa-4c39-8685-7307866c0fe0.mp4

I also have a longer video of it happening while moving forward.

[vshcherb]

Looked at short video, some things need to be clarified:

1. Is it the latest 4.3 version?
2. Car was moving in opposite direction than suggested route?
3. What was map orientation mode?

It's also not clear whether the car was moving or standing still. We have quite a big refactoring how to handle when the car stands still, so we disabled map rotation to compass which was enabled with low speed.

[sonora]

Also: We have the Kalman filter (Menu button > Configure profile > General settings > Use Kalman filter) to mitigate such situations. But from the look of things it may already be activated?

[ZeLonewolf]

Here is a longer video of it happening: https://youtu.be/BJ66l2WTtA4

In both videos, I'm driving along the planned route, but the orientation is flipped from the expected. In the short video, I'm stopped at a red light and in the video here I'm driving down the road.

Version info

Driving mode settings

[vshcherb]

It looks totally wrong indeed it makes sense to try switch other settings but you need to do it in a car profile

• Turn off "Animate own position"
• Turn off "Use Kalman filter"
• Make sure "Map orientation mode" set to "Movement direction"

[sonora]

Could also be the device orientation is detected wrong for this setup: Test the fixed device orientations "Portrait" and "Landscape" rather than "Same as device" under Menu button > Settings > Screen orientation (this is a different setting than the Map orientation).

(We once had a support ticket with a rather non-standard hardware (a camera) where the device x-, y-, and z-axis were left-handed.)

[Zirochkabila]

OsmAnd~ 4.4.0#14059mqta, released: 2022-12-29 @ZeLonewolf As you can see in the video - it could not be reproduced. Try to set the profile settings to default: Menu > Configure profile > Reset to default or do as wrote about it @vshcherb and @sonora

https://user-images.githubusercontent.com/104760013/209995953-2709a368-103c-4095-afde-026596732750.mov

[ZeLonewolf]

Thanks - I will run some more tests with the settings described here once I'm back from vacation.

[nicosensei]

I'm having the same issue, reached out to support and was directed to this issue ticket.

I'll try the workarounds but I have little confidence it will work, because car orientation is properly detected on the app, i.e. this issue happens only when connected to my car through Android Auto.

Here's my device I fo:

Device : vayu Brand : POCO Model : M2102J20SG Product : vayu_global Build : SKQ1.211006.001 release-keys Version : 12 App Version : OsmAnd+ Apk Version : 4.3.4 4304

[sonora]

Watching the available movies, it seems to me the issue occurs only if you do not follow the calculated route but instead go the opposite direction? Or is that an artifact, and the blue arrow is actually pointing wrong?

Also: It somehow reminds of #16081, have you tried with both "Animation" settings disabled?

[ZeLonewolf]

It occurs when following the calculated route, and the blue arrow is point wrong. I'll try to make a longer video today.

[sonora]

Has anybody tried my hint to use the fixed device orientations "Portrait" and "Landscape" rather than "Same as device" under Menu button > Settings > Screen orientation (this is a different setting than the Map orientation)?

[ZeLonewolf]

Has anybody tried my hint to use the fixed device orientations "Portrait" and "Landscape" rather than "Same as device" under Menu button > Settings > Screen orientation (this is a different setting than the Map orientation)?

I've tried changing the device to used the fixed orientation mode of "landscape". This changes the orientation on the phone but does not appear to impact the behavior on the auto display.

[ZeLonewolf]

Sorry for the delay, below are some video sequences that I captured earlier in the month.

Jan 3: (default settings) https://www.youtube.com/watch?v=lH4niDcWVnc https://www.youtube.com/watch?v=I8tiHduXsIc https://www.youtube.com/watch?v=43se_kzs5hw https://www.youtube.com/watch?v=Qy__8fhHkYs

Jan 4: (landscape mode enabled) https://www.youtube.com/watch?v=PIqZW6J37kM

[sonora]

It reminds me of this story #2131, where we had the weird occurrence that the device in question (in your case the car screen) had a non-standard internal coordinate system and hence OsmAnd interpreted some orientation-related directions falsely.

In e.g. video 1 it seems that the left hand edge of the car screen is assumed to be the top edge, therefore the blue position arrow always points towards it (in orientation mode "Movement direction" the arrow always point towards the top of the device). But the map, at the same time, seems to follow a different rotation logic, that's why the blue arrow direction contradicts with the underlying street orientation.

"Sign error", I would say. Unfortunately I would have a hard time debugging because I have no way of testing AA, so I am hoping someone else can take a look.

BTW: Maybe I missed your response: "have you tried with both "Animation" settings disabled?", just to check if this is in the animation or more fundamental code portion.

[ZeLonewolf]

I'm planning to run a test this evening.

I see two settings:

• No animations (currently OFF)
• Animate own position (currently ON)

I guess it makes sense to try one at a time?

[sonora]

Yes, exactly thanks! Anything would help to locate which code portion would need checking.

Also: I understand the issue is limited to using Map orientation= Movement direction? If you set it to e.g. "Fixed (North is up)", you see no such anomalies?

[ZeLonewolf]

If you set it to e.g. "Fixed (North is up)", you see no such anomalies?

That's a good question. I would never navigate with the map set to fixed north! :sweat_smile: But I can certainly try it in the interest of science.

[Zirochkabila]

OsmAnd~ 4.4.0#14092mqta, released: 2023-01-13 We managed to reproduce a similar behavior when Simulating location by GPX (navigation didn't start)

https://user-images.githubusercontent.com/104760013/212315637-82c2e1b5-2d69-4b79-9fc1-4d3039d3a826.mov

[vshcherb]

Track doesn't have heading / bearing provided so it looks like we could "auto generate " heading in this case.

[sonora]

@vshcherb Do I understand it correctly that the original issue reported here is not a general issue affecting all users, but is specific to this particluar hardware setup? In other words, the vast majority of users use Androind Auto in mode "Movement orientation" without such issue, and as a consequence we are not looking for a general glitch in our code, but rather for something which only occurs in certain environments or under certain circumstances?

[vshcherb]

it's not clear yet but very likely it's a rare case & device specific.

[ZeLonewolf]

Here is a video I took a few days ago:

https://www.youtube.com/watch?v=apUTdMWmsWw

For this, I used the settings:

• Animate own position: OFF
• No animations: OFF

[ZeLonewolf]

Note that in my original tests, the behavior in https://github.com/osmandapp/OsmAnd/issues/16041#issuecomment-1381759456 was exactly what I saw.

[sonora]

Thanks for the new video! We can conclude that

• This seems not related to the animation settings
• In all videos the map orientation is wrong. (It should always be such that the current direction of the calculated route as displayed points towards "up".)
• In all videos the direction the blue position arrow points is wrong, but differently wrong. It should also always point up, but points in a fairly consistent but arbitrary direction with constant small-angle fluctuations.

It does not look like merely a false coordinate system.... more like the magnetic sensor or some similar slow-changing effect is in play here... or a bearing value is submitted which confuses the output.

[alex-osm]

There is a topic at Android Auto Help Center. May be this issue is related to AA itself. I would suggest to update Android Auto app and Google Play Services on the phone and probably update car multimedia system as well.

[alex-osm]

Other thing is try to change Location Source

[ZeLonewolf]

That linked Android Auto issue sounds different from what I'm experiencing - the position does not fly around to random locations, and Google Maps navigation works normally with no artifacts.

[sonora]

@vshcherb I don't understand why we still have one open Issue each in the milestones 4.3-android and 4.3-backend, which are both closed? An oversight? Shouldn't the Issues be transferred to 4.4?

[vshcherb]

nope yet, 4.3.9 still be done, that's under control.

[vshcherb]

Setting was added (we're discussing whether it's advanced development option or should be default or should be device dependent)

[ZeLonewolf]

Not sure I understand the final disposition - this bug will occur by default unless a special setting is enabled?

[vshcherb]

With this setting enabled you should not have bug with bearing incorrectly produced by device

[dmpr0]

Doc team We need to mention in Android Auto article about new option Approximate bearing

• Visible only if OsmAnd Development is enabled
• Placed in Menu – Settings – Profile – Navigation settings – Map during navigation