I am trying to track the position of the ADIS16480 sensor by double integrating the accelerometer values. The ADIS16480 is meant to run in inertial frame so that the position is output in reference to ENU.
I set up a testbed to test the integration algorithm along one axis, put the system into local navigation frame (ENU), set a decimation rate of 8 (sample rate is therefore 270Hz) and let it TARE itself.
So EKF_CNFG = 0x200
DEC_RATE=0x8
Then I moved the ADIS along the x-axis.
Raw results of the x and y axis look like this (y - axis in g, x-axis in samples):
Up to about 800 samples the plot shows what I would expect. However after the movement has stopped (around 900 samples) the accelerometer values return to 0 in a slope. I have seen this behaviour across a number of experiments and DEC_RATE settings, but they tend to disappear when EKF_CNFG = 0x208 (body frame).
Can somebody confirm this behaviour?
Thanks a lot,
Rapha