Multiple Fading Factors Based Adaptive SVD-aided UKF for Small Satellite Attitude Estimation

This paper discusses how to modify the system noise covariance matrix for a nontraditional attitude estimation filtering technique. The nontraditional attitude estimation approach combines the Unscented Kalman Filter (UKF) and Singular Value Decomposition (SVD) method to estimate a nanosatellite's attitude. As the algorithm's initial stage, the SVD technique calculates the nanosatellite's attitude and provides one estimate at a single frame using readings from the magnetometer and Sun sensor. Then, an adaptive UKF is fed with these estimated attitude terms. The attitude estimation of the satellite are compared between the UKF and the suggested adaptive UKF. It is suggested to adapt the Q (system noise covariance) approach using various scale factors. In the event of an increase in process noise caused which could be induced via modifications to the satellite's dynamics or the surroundings (space environment changes e.g. eclipse period), performance of the multiple scale factors based adaptive SVD-Aided UKF algorithm is investigated.