Using A Gimbal To Calibrate Inertial Sensors

Abstract

A novel technique for calibrating the orientation, position, scaling and offset of each of the sensors in an inertial measurement unit (IMU) is presented. An IMU consisting of 3 accelerometers and 3 rate gyros is created using off-the-shelf sensors. A gimbal is designed consisting of three concentric rings, with rotary encoders measuring the rotation between rings. The IMU is fixed to the inner ring of the gimbal and rotated in space. By sweeping appropriate orientations of the IMU at appropriate rates, filtered sensor values can be mapped to true angular velocities and linear accelerations computed from the gimbal rotations. The sensor parameters are estimated using Minimum Mean Squared Estimation (MMSE), and a Kalman filter is implemented to estimate the IMU's attitude (roll and pitch angles) from the raw sensor values. The calibrated sensors are found to track the pitch angle with a RMS error of 1.74 degrees, and the roll angle with a RMS error of 3.14 degrees. The novel outcome of this research is that it defines a technique for calibrating IMUs with component sensors that need not be orthogonal in placement. - inertial measurement units, calibration, gimbal