Satellite attitude estimation based on marginalized particle filter

An algorithm based on marginalized particle filters (MPF) is presented to solve satellite attitude and gyro bias estimation problem with vector observations. By marginalizing out the state appearing linearly in satellite model, attitude vector is approximated by a set of particles and estimated using particle filter, while estimation of gyro bias is obtained for each one of attitude particles by applying the Kalman filter, which is associated with each particle in order to reduce the size of the state space and computational burden. The method of estimation with equation constraint is employed due to the normality constraint of attitude quaternion. The numerical simulation of a rigid satellite with gyro and three-axis-magnetometers, shows the superiorty of the proposed algorithm in coping with the nonlinearity of model.