I’m working on a system to create an IMU temperature calibration fit simply from flight data spread out over different days and different conditions. Here I am testing the initial filter convergence rate with an autonomous take off strategy that depends on accurate pitch and roll estimates right from the start.
How did it work? Do you plan to post some updates with screenshots of the telemetry/IMU data processing? I’d be interested in seeing to what extent you were able to measure performance improvements with the temperature compensation.
It’s definitely a work in progress, but I do plan to continue to share my results. I working on a number of fronts … autopilot code, sensor firmware code, possibly designing a board, working with a 15-state EKF developed by a professor at the U of MN. I’m actually really thrilled with the calibration I can get from the flightdata, it seems to agree pretty much with what you get from a room-temp ground calibration, but I can extend the fit over the temperature ranges that were observed during the flights. And there is a fair amount of change in the bias as the electronics warm up or the air/cabin temperature changes.