On-Line Orientation Calibration of Inertial Measurement Unit Pairs using Unmanned Aerial Vehicle Flight Data

Abstract

Calibration of the orientation of multiple Inertial Measurement Unit (IMU) pairs was performed using flight data collected from six separate flights by the West Virginia University (WVU) ‘Red Phastball’ Unmanned Aerial Vehicle (UAV). Four closely grouped IMUs were located in the front of the UAV for all six flights, and two sets of flight data featured a fifth IMU positioned at the rear of the UAV. This enabled calibration of IMU pairs with negligible distance as well as calibration of spatially separated IMUs, which were modeled assuming rigid body mechanics. Accelerometer and rate gyroscope data were used to produce two linearly independent vectors which were used to define the relative orientation of each IMU pair via an Unscented Kalman Filter (UKF). Using static UAV data recorded before each flight, the gravity vector was used for off-line calibration of the pitch and roll of each IMU with respect to the Earth reference frame, which was used as a truth reference to evaluate the accuracy of the on-line calibration techniques.