Noisy Acceleration in FullMotionCalibrarion

[milopeters]

Acceleration data generated by the FullMotionCalibration opMode is, in contrast to the velocity data, very noisy and unusable. This may be caused by instability in dt, although this should correct any noise, instead of causing it. Currently, max velocity data is usable for motion profiling, but acceleration is not, as it sometimes spikes to impossible values, and sometimes reports a max acceleration that is obviously too small (eg 100 cm/s² when it accelerates to 140 cm/s in under a second).

[Braedeng0]

A Kalman Filter can be used to estimate acceleration based on measured velocity data and should provide a less noisy result.

[milopeters]

I also think that a lot of the noise is coming from the velocity, which when differentiated makes the acceleration very noisy. I could also add a low pass filter when plugging in for acceleration.