Square root unscented Kalman filter based on strong tracking

Abstract

To solve the numerical instability in the recursive process of unscented Kalman filter (UKF), as well as the unsatisfactory performance in case of abrupt changes, a new adaptive target tracking method, called square root unscented Kalman filter based on strong tracking (STF–SRUKF), is presented. On the one hand, inspired by the idea of square-root filter, the square root of the covariance matrix is substituted for the covariance matrix itself in the recursive process, to guarantee numerical stability. On the other hand, based on the idea of strong tracking filter, a time-varied fading factor is introduced into the recursive process, which is helpful to adjust the gain matrix timely, and thus enabling STF–SRUKF more power to deal with sudden changes. Experimental results demonstrate that STF–SRUKF performs well and steadily, especially when target motion changes suddenly.