Attitude filtering for nanosatellites: A comparison of different approaches under model uncertainties

Abstract

Abstract Nanosatellites have become an integral part of space research in recent years, especially for universities and non-profit organizations. Attitude estimation is a key function that is playing a leading role in the success of their missions. Despite various research and several implemented algorithms, it is difficult to find comprehensive analyses on the attitude filtering algorithms used for nanosatellite attitude estimation. This article, first of all, presents different attitude filtering approaches that can be used for estimating the attitude of a nanosatellite. These approaches include the Multiplicative Extended Kalman Filter and Unscented Attitude Filter algorithms and their versions integrated with the QUEST algorithm that are pre-processing the vector measurements. Then the given filtering algorithms are evaluated with their different structures depending on the availability of gyro measurements; gyro-based filtering, gyroless filtering and dynamics-based filtering with gyro measurements. These evaluations are done with a realistic dynamics model that includes the dominating disturbance torques for a nanosatellite mission which are assumed unknown at the stage of attitude filtering. In the end, twelve different filtering algorithms are compared with Monte Carlo analyses.