Spacecraft attitude estimation from vector measurements using particle filter integrated with unscented Kalman filter

A particle filter (PF) based algorithm is given to solve the attitude estimation problem of the three-axis stabilized based on vector observations and biased gyro. In order to avoid the potential computational burden problem associated with the number of required particles, this algorithm using a dual filter; integrating the PF with the unscented Kalman filter (UKF) is employed, in which the PF is used to estimate the quaternion and the latter to determine the gyro drift errors. The attitude is expressed by three-component vector generalized Rodrigues parameters (GRPs), where only three parameters are needed to describe orientation and the singularity of the covariance matrix is also avoided when unit quaternion is used in the attitude estimation due to unit norm constraints. The efficiency of the PF estimator is tested through numerical simulation of a fully actuated rigid body with gyro and three-axis-magnetometers (TAM). For comparison, UKF estimators are used to gauge the performance of PF estimator. The results clearly demonstrate that the PF is superior to UKF in coping with the nonlinearity of attitude dynamics.