Visual approach for MEMS gyroscope drift compensation

This paper presents the design of an attitude determination system from a small satellite perspective that automatically corrects for attitude drift using relative attitude measurements generated using a small imager. The visual attitude propagation approach, referred to as the 'stellar gyroscope', is a star imaging system that generates relative attitude measurements between star field images with common stars, without using a star database. The Random Sample Consensus (RANSAC) approach used in star pair detection offers significant noise immunity, which allows the approach to be implemented using miniature cameras with lower signal to noise ratios, providing a new alternative for miniature spacecraft (such as CubeSats). The various mechanism of attitude estimate drift when using MEMS rate gyroscopes are discussed in detail. The stellar gyroscope concept is described, and a Kalman filter is described that can utilize the relative attitude measurements provided by the stellar gyroscope in order to maintain a high accuracy attitude fix throughout the orbit.