Ellipsoidal set based tracking with nonlinear measurements

There are many target tracking applications where measurements are nonlinearly related to target state. Typical examples of this include active sonar tracking with range-bearing measurements and passive sonar tracking with bearings-only measurements, where target state is position and velocity in Cartesian coordinates. In this paper, ellipsoidal set based estimators are derived for both of these nonlinear measurement scenarios. The premise of an estimate that provides a guaranteed bound on target state is maintained throughout the paper. Monte Carlo simulation results are presented for each type of nonlinear measurement, including a comparison with the extended Kalman filter in the case of bearings-only measurements. © Australian Crown copyright, 2005.