A Factor Graph Approach for Attitude Estimation of Spacecraft Using a Stellar Gyroscope.

A stellar gyroscope is an attitude sensor that is composed of a micro star sensor and a micro-electro-mechanical system gyroscope. The attitude is the output by fusing the information of the two sensors. Due to the processing of the star image from the star sensor, a measurement delay is introduced and can affect the performance of the fusion of the two sensors. In this paper, a factor graph method that is based on a sliding window is investigated for dealing with the measurement delay. Temporary factors are calculated to temporarily occupy the places of the factors that are under processing, and sliding window incremental smoothing is only performed with a set of factors and states after the delayed measurement has arrived. To reduce the frequency of attitude calculation based on a star image, the keyframe strategy is adopted. Only in keyframes, the attitude is solved by using the star image and used as the measurement of the factor. In non-keyframes, only the star positions in the star images are determined, with a robust star tracking method that is aided by the gyroscope. Moreover, a high-rate low-latency output framework that is by a gyroscope is proposed for accelerating the attitude output rate. According to the results of the experiments on simulated data, the proposed method can overcome the problem of measurement delay and is more effective than the compared methods. According to the results of the experiments on real data, the proposed method can be used in applications.