UKF and Its Application to Bearings-Only Tracking Problem
Shu Wang · Flight Dynamics · 2003
An application of the unscented Kalman filter (UKF) to the two dimensional bearings only tracking (BOT) problem in passive target tracking from a ground station is presented. The target position and velocity estimates are approximated by a Gaussian distribution which is specified by a set of deterministically chosen sample points. At each update, the sample points are propagated through the state equation and then transformed through the nonlinear bearings measurement equation. From these sample points, the posterior mean and covariance of the target position and velocity are computed accurately to the second order. The linearization of the nonlinear equations necessary for the extended Kalman filter( EKF) is not needed. The simulation results show that in the BOT problem this UKF outperforms the standard EKF in accuracy and divergence performance.