What is the best approach to calculate the average angular velocity in the global frame when the sensors are mounted off the X, Y, and Z axis? I believe I see a few possibilities, such as averaging yaw (fPsiPl) over time or using the average angular velocity (fOmega) along with the roll (fPhiPl) and pitch (fThePl) using trigonometry. I'm looking for good sensitivity. The average angular velocity works well set to ±250dps and referenced to an axis. Now I want to prove out off axis velocities. I'm using the 9DOF Kalman filter with SensorFusion 5.
Secondly, does the average angular velocity gyro offset errors? I'm seeing some small amount of bias in each axis. Should I be subtracting the average of each bias out? I'm trying to sense velocities of < 6 deg/s.
Thanks
Jeremiah,
Apparently a couple of my email replies to the platform bounced back. Your problem sounds like a basic trig orientation computation. I'll post my "Frames of Reference" cheat sheet to the community. Take a look and see if that helps. If not, we can set up a call. You can reach me "off-community" at mike.stanley@nxp.com.
Regards,
Mike
Jeremiah,
Do you mean off-center, but axis-aligned? Or actually rotated from the reference axes? In the first case, simply average the gyro-offset-corrected angular rates. In the second, average and then rotate the averaged values.
MIKE
Mike,
Thanks for the info. I mean actually rotated from the reference axes. I need to be able to detect angular rotation in the yaw plane (fPsiPl) of the global frame. Hopefully I can resolve the sensitivity back to what is was on axis for small angular velocities.
Jeremiah