Bearing Only Tracking for Maneuvering Target: EKF, UKF and CKF

The bearings-only tracking (BOT) is a nonlinear estimation process that determines the moving of the target based on a time series of bearing measurements from a single or more observers. It is assumed that the dynamic of the target is constrained to a straight line, constant speed segments separated by maneuvers in position and speed. The extended Kalman filter (EKF) is often used in BOT and has good computational efficiency. Unfortunately, this class of filters often leads to unstable estimations. The unscented Kalman filtering (UKF) and cubature Kalman filter (CKF) have recently been suggested for stability improvements, and UKF is suggested to be more suitable for real-time applications than EKF. In this paper, a comparative analysis between the EKF, UKF, and CKF implementations is provided.