A back-propagation approach to compensate velocity and position errors in an integrated inertial/celestial navigation system using unscented Kalman filter

Abstract

This article aims to compensate the velocity and position errors that exist when the star sensor starts to work in a strapdown inertial navigation system aided by celestial navigation. These systems are integrated via unscented Kalman filter to estimate the current attitude and the gyros fixed bias, precisely. Since an accurate integration is desired, the nonlinear attitude equations are utilized in filter and these equations are propagated through a precise discretization method. Then, implementing the back-propagation and smoothing techniques, the initial attitude and the accelerometers fixed bias are also estimated. Finally, carrying out a parallel navigation, the velocity and position errors are compensated. The validity of the proposed method is investigated through simulation of launch vehicle navigation. Simulation results show a great reduction in velocity and position errors.