Autonomous satellite navigation using landmarks

Covariance analysis is used to assess the ability of a satellite to autonomously determine its orbit and attitude and the position of features of interest observed by an earth sensor. The measurements employed are angular rate data from a three-axis inertial reference unit, the positions of known landmarks viewed by the earth sensor, and some combination of star and sun data. The 12x12 covariance matrix contains six orbit elements errors, three attitude error angles, and three gyro drift rate errors. A closed form approximation to the process noise covariance matrix is derived for the assumed Gaussian process noise sources: unmodeled intrack, radial, and orbit normal accelerations, gyro output noise, and gyro drift rate derivative noise. Test runs indicate that the intract orbit uncertainty is the major component of the geopositioning error.