I'm considering upgrading the IMU in the odometer of my wheeled robot, potentially from ICM-20608 (and MPU6500 previously) to ADIS16505-1, as I heard that ADIS IMUs are better than ICM/MPU series. Therefore, I'm investigating why I should do this and how much more accuracy I could expect.
However, when I did an experiment to estimate the in-run stability of ICM-20608 using the procedure posted here [https://ez.analog.com/mems/w/documents/4510/faq-gyroscope-in-run-bias-stability], I got 5.33 deg/h, which is comparable to ADIS series.
I know that some ADIS IMUs are well-calibrated, especially regarding axis-axis alignment. But for an indoor wheeled robot, I actually just need the yaw gyro, which means the such calibration result is not that critical in this case.
ADIS IMUs also has lower drift over temperature change, but we have setup a temperature control unit on the odometer, so temperature change is not a critical issue any more.
With the above facts being mentioned, I'm beginning to wonder how much ADIS IMU could help increase the yaw estimation accuracy in my robot. Do you have any insights on this? Thank you!
Other than axis-axis alignment, sensitivity error (ADIS16505 (Rev.A datasheet, page 3): +/-0.5%.typ. -40 to 85C vs ICM-20608-G (Rev 1.0 datasheet, page 7): +/-2% @25C.tp. and +/-3%.typ. @-40 to 85C…
Other than axis-axis alignment, sensitivity error (ADIS16505 (Rev.A datasheet, page 3): +/-0.5%.typ. -40 to 85C vs ICM-20608-G (Rev 1.0 datasheet, page 7): +/-2% @25C.tp. and +/-3%.typ. @-40 to 85C) might affect to the odometry especially during every turn, unless you don't calibrate the sensitivity. Another point is gyro offset error due to vibration (linear vibration effect and VRE, see page 3 of ADIS16505 datasheet), which is not mentioned in ICM-20608-G datasheet but might affect to odometry especially when there are shock and/or vibration due to bumps on the road. Hope this helps!