High-degree cubature Kalman filter and its application in target tracking

A high-order cubature Kalman filter (HCKF) based on the arbitrary-order cubature rule was proposed and applied to the maneuvering target tracking problem resulting from the limited precision of the conventional cubature Kalman filter (CKF). The conventional CKF, which employs the third-order spherical-radial cubature rule, can achieve better estimation precision and numerical stability than the other nonlinear filters, such as Unscented Kalman Filter(UKF). To further improve the estimation precision of CKF, in the framework of point-based Gaussian approximation filters, Genz's method and the moment-matching method were employed to deduce the arbitrary-order spherical rule and the radial rule, respectively. By combining the two rules, the high-order spherical-radial cubature rule was developed for computing Gaussian weighted integrals, and HCKF was proposed based on the high-order spherical-radial cubature rule. The proposed HCKF algorithm was applied to the maneuvering target tracking to test its performance. The simulation result shows that HCKF can achieve better estimation precision than conventional CKF, with the improvement of position and velocity estimates by 11% and 24% respectively.