Positional Tracking - INERTIAL

SDK 5.0.2
have tested with Positional Tracking GEN1, GEN2 and GEN3
ZEDX camera with SVO recording.

Camera and Jetson mounted to Vehicle

We noticed during testing that if the camera goes into a dark room for maybe 10 seconds the positional tracking (spatial mapping status) transitions to INERTIAL however the X,Y,Z translation does not update when in this mode, the Rotation does but its sometimes a little bit off… Should the camera be able to perform dead-reckoning even for a short moment when it has no visual? i was under the impression that is what it did hence transitioning to using only INERTIAL measurements.

in the PositionalTrackingParameters I have set enableIMUFusion = true

One other thing we noticed is with GEN2 once it does gain visual again it takes around 500+ frames for it to transition to VISUAL_INERTIAL… this is also problematic because if we do enter a dark room but happen to gain VISUAL for a moment then it wont transition back to INERTIAL and tracking is lost entirely.

Any help would be greatly appreciated

Regards,
David

Hi @dhutchinson
Welcome to the StereoLabs community.

GEN_1 and GEN_2 use the IMU information only to improve the rotation estimation.
So if visual information is lacking, only rotation is estimated.

GEN_3 is in continuous evolution. It currently uses IMU in the same way as GEN_1 and GEN_2, but we are working on improving this.