Hi all,
I'm working on integrating a BMI323 IMU for an application requiring continuous measurement of roll/pitch using a complementary filter and am observing strange behavior I'm not sure how to explain.
Essentially, translating the sensor in a conical fashion without any rotation in the third sensing axis will result in an angular rate nonetheless being measured on this third fixed axis. The attached image should help visualize the motion described. Essentially, imagine a line of fixed length drawn between two points in space - one end of the line remains fixed in place (tip of the cone), and the other with the IMU is free to translate in a circular motion (base of the cone). In the attached image, this would result in the IMU experiencing both a pitch and a yaw component to make up the circular translation/motion. However, even if no roll is imparted upon the IMU (dot remains "up", in the -Z direction) the IMU measures a non-zero angular rate for roll. More interestingly, if the direction of travel around the cone's base is reversed, the sign of the measured roll angular rate will also reverse. The same behavior is observed if the sensor is moved in a conical shape about the y axis (roll and yaw are varied, resulting in a measured pitch angular rate despite no change in pitch).
Based on these observations, I have a few questions:
Is this behavior expected for the BMI323 specifically, potentially due to how the microstructure is designed? I did note cross-axis sensitivity, which this perhaps could be related to, is not specified in the datasheet for this component as it is for other parts.
Alternatively, is this more broadly a physics phenomenon at work that would impact all gyros? If so, what is the name for this effect and are all gyros equally susceptible?
If this behavior is entirely unexpected, I will further review the complementary filter but so far it has performed exactly as expected in all other cases.
Thanks!
