Non-linear robust filter design for satellite attitude determination

In this study, a non-linear robust filter (NRF) is proposed for a satellite attitude determination system. The system under consideration is subject to both the model parameter uncertainties and the linearisation errors. The proposed NRF is derived to guarantee an optimised upper bound on the estimation error covariance. The characteristics of the filter are analysed using some standard results from stochastic stability theory. It is specified that the stability of the proposed algorithm can be guaranteed independent of the model uncertainties as well as the linearisation errors. In addition, the efficiency of the NRF is demonstrated with the telemetry data from a satellite. The algorithm is developed in general such that it can be applied to stochastic uncertain non-linear systems other than satellite attitude determination.