strange accelerometer measurements of Microstrain MEMS IMU on an octocopter

[JzHuai0108]

Hello,

I tested the sensor_fusion framework with only IMU measurements which were collected by an Microstrain MEMS IMU mounted on an octocopter. The orientation results looked fine. But absurdly, the position in the vertical direction got negative, that is, the UAV was flying underground.

When the accelerometer readings were drawn in attached graphs, I was confused by accelerometer readings at z-axis.

The IMU was mounted on the octocopter with z axis pointing down, so the accelerometer on z axis should sense about -9.8 m/s^2 when the UAV is in smooth flight. Even it inclines somewhat, assuming the maximum inclination is 15 degrees, the accelerometer on z axis should still sense about -9.8*cos(15 degree)=-9.47 m/s^2. However, as shown by the second graph, it only senses around -8 m/s^2 on average. As a result, the IMU seems accelerating into the underground. To make it fly up, a bias of 1.6 m/s^2 has to be applied.

However, based on our experience and its specs (http://files.microstrain.com/3DM-GX3-35-Attitude-Heading-Reference-System-GPS-Data-Sheet.pdf) , the expected accelerometer bias should be less than 4 mG.

Do you have any idea what caused this strange behavior of IMU on octocopter? Maybe vibration?

Thanks for your consideration,
Janju Huai