@legatus Hi thanks for the mention, in fact I studied his paper and it turns out to be exactly what I need. Although he only conducted the experiment for yaw angles, I would also consider taking roll and pitch angles into consideration (not all 360 degree since I am using a drone as well). In practice, the drone will move whereas there are roll/pitch/yaw movements at the same time. Am I right to assume that if I know the offset (and its variance/certainty) for a specific orientation, I can interpolate the final offset from the single yaw/pitch/roll offset?
Probably going to use my VICON system for ground truth data instead of going for IMU fusion like he does.