UKF based on-orbit gyro and magnetometer bias estimation as a part of the attitude determination procedure for a small satellite

In this study an unscented Kalman filter (UKF) based magnetometer and gyro bias estimation algorithm is presented as a part of the attitude determination procedure for a small satellite, which has three magnetometers and three gyros as measurement sensors. Differently from the existing studies both magnetometer and gyro biases are estimated as well as the attitude parameters. Kinematics of the satellite is modeled via using quaternion parameterization. In order to enable estimation of all parameters simultaneously without a high computational cost, two methods known in advance are used. First, quaternion based modeling constraint that is met because of the unit norm of quaternion is overrun by enforcing constraint as a dummy measurement through the measurement model, instead of employing three-element error quaternion state as attitude representation. Second, in filter formulations gyro-based model is used for attitude propagation. Validity of the proposed method is confirmed by simulations.