Tracking a target using a cubature Kalman filter versus unbiased converted measurements

In tracking applications, the target dynamics are usually modeled using Cartesian coordinates, while the measurements obtained by a sensor are reported in polar coordinates. In this case, there are four filters for the target tracking: the Kalman filter with unbiased converted measurements (UCMKF), the extended Kalman filter (EKF), the unscented Kalman filter (UKF), and the cubature Kalman filter (CKF). A comparison of the UCMKF with the EKF shows that the UCMKF provides better estimation accuracy than the EKF, while the comparisons of the EKF, the UKF and the CKF show that the CKF provides the best performance for the target tracking among them. The UCMKF or the CKF, which one is better in the performance is a problem to be researched. To do this, a CKF for a nonlinear observation is proposed in which the three-degree spherical-radial rule is applied to solving the nonlinearity in the observation equation. The performance comparison between the UCMKF and the CKF has been done by simulations, which shows that the CKF provides better tracking performance than the UCMKF.