STRENGTHENED CONVERGENCE OF ESTIMATIONS IN THE IN-FLIGHT GEOMETRIC CALIBRATION

The problem of in-flight geometric calibration of the spacecraft's imaging complex is considered. Here calibration is interpreted as making the more precise mutual attitude of the onboard imaging camera and star tracker in a spacecraft's body. It is a necessary part of preparing the optical-electronic complex for imaging and geo-referencing of the ground objects. Received snapshots and readings of star tracker and GPS are processed on the ground. In calibration, snapshots of landmarks are used. There exist methods of in-flight geometric calibration on the base of obtained measuring equations of various physical origin. These equations are solved by means of the least square method. Usually, snapshots of known landmarks are used, but the problem's solutions with attracting of unknown landmarks are possible. In this work, an approach to the solution of the in-flight geometric calibration problem using formulas of a fuzzy observer is proposed. In many cases, such an approach allows diluting the negative influence of disturbances and sensor errors onto the accuracy of the estimation of calibration parameters. Two versions of a fuzzy observer for the in-flight geometric calibration are elaborated. The first one takes into account the whole manifold of the obtained snapshots at once. The second version of the observer has recursive character. All snapshots are processed successively one after another with immediate correction of the calibration parameters. Such an approach allows improving convergence of estimates. As in such a case, estimated parameters of calibration are constant, a stage of prognosis peculiar for such algorithms is not necessary, and only update procedure is used. Narration and arguments are accompanied by sufficient volume of computer simulation with the use of known or unknown landmarks, in particular, in conditions of an anomalous initial discrepancy. The results of simulation confirm above-mentioned advantages of two versions of the fuzzy observer as compared with the least square method for in-flight geometric calibration.