Vuforia Model Tracking crashes ML2 - rebooting the whole device

I got Vuforia Model tracking to work on the ML2 and now I am experimenting with different model/marker ideas.
One thing I did was to 3d-scan an object and then created an advanced model target from it. It actually works - the object is being identified correctly and then tracked - for a few seconds.
After about 15-20 seconds, the whole headset shuts down and reboots (Magic Leap loading bar etc.)

The app reports this as an error before shutdown:
Autoconnected Player "Autoconnected Player" Vuforia Engine has stopped due to an error: The operating system has dropped the camera device used by Vuforia Engine

I'm running the following specs:
ML OS: 1.3.1 (found 1.4.0 to be highly unstable, had crashes/reboots of the device every 15-20 minutes)
ML SDK: 1.9.0
Vuforia Engine: 10.17.4
Unity: 2022.3.5f1

Hi, it sounds like it might be a device or OS issue. Can you confirm this is happening only with Vuforia and no other app that uses the camera?

@Tom_LEFX can you please pull and send us logs from the device?
@etucker can help with that if you do not know how.

Thanks for getting back to me and sorry for the delayed response. It has been a bit busy these days. I'll check the application again today and will try to pull logs after it crashed the device.

Do I need to do something in particular to get logs from the device after a full crash? Or is it just a continuous logfile anyway?

So far, it only happened with the Vuforia Model tracking. Vuforia image and marker track ran fine for longer time. I also tried ML Assist and it did not crash. (at least on 1.3.1 OS - 1.4.0dev1 crashes are frequent for me with many apps)

I wanted to reproduce the issues today - but suddenly the app just does not want to crash the HMD anymore. Same app, same OS - only difference: It's about 7-8 degree colder in the room than last week. Could it have been a thermal issue? I hear (from the fan noise) that the model tracking seems to be quite compute intensive.