Augmented Robust Cubature Kalman Filter Applied in Re-Entry Vehicle Tracking

Abstract

In this work, an improved cubature Kalman filter (CKF), called augmented robust CKF (ARCKF), for the reentry vehicle trajectory tracking is presented, in which the model strong nonlinearity, heavy-tailed measurement noise, and the non-independence of measurement noise in iterations are considered. Firstly, the derivative-free robust approach is employed instead of the conventional Huber's technique, thereby eliminating linearization errors and exhibiting robustness to measurement outliers. Secondly, the framework of iterated unscented Kalman filter is adopted to avoid linearization in the iterative measurement update, in which new cubature sample points are regenerated in each iteration, and then the probability density function is propagated through the measurement model. Furthermore, in the measurement update stage, the state vector is augmented with the measurement noise vector to address their correlation after the first iteration. To demonstrate the validity of the proposed algorithm, CKF, robust CKF (RCKF), and ARCKF are compared via Monte Carlo simulations. The results show that ARCKF performs with sufficient suitability in non-Gaussian noise environments and is superior in terms of target tracking accuracy.