Fusion pose from previous frames

Here is my goal:

Given a fusion instance that is actively calibrating a camera’s world pose from GNSS data, get the pose transform from a previous timestamp.

My thinking is that the pose from a previous position/timestamp might become more accurate after GNSS calibration that in a given moment.
(pose corresponding to time 0:01:01 would be more accurate after several GNSS ingestions)

Is this possible with the current python SDK?


This is not implemented “as is” in the SDK. Theoretically once you get a proper GNSS calibration you can always apply the previous non-fused localization path in reverse to find the previous poses, but I’m not sure that is what you need.

Be aware that you can enable (or not) the rolling calibration. With it, you’ll get values immediatly, without you’ll have to wait until you get a proper enough calibration.

Where in the fusion initialization do I set it to rolling calibration?


Hi, you do that from the GNSSCalibrationParameters: GNSSCalibrationParameters Class Reference | API Reference | Stereolabs