Use of Target-Oriented Process Noise in Tracking Maneuvering Targets

摘要

Kalman filters for target tracking typically use a process noise defined in thetracking frame. This requires setting the variances of the process noise in all three Cartesian coordinates to values large enough to compensate for the largest possible target maneuver. In general, the maneuvers performed by aircraft flying at high speeds are constant speed turns in a plane which are known as coordinated turns. An aircraft performs a coordinated turn by controlling its lateral acceleration which is approximately orthogonal to its velocity vector. Thus, it is logical that most of the uncertainty involved in tracking these targets is in the direction of the lateral acceleration. This information can be exploited by supplementing a standard Kalman filter with a target-oriented process noise. A target-oriented process noise is obtained by rotating a process noise, defined in the target's frame, into the tracking frame. The rotation matrix is computed from the state estimates of the Kalman filter. The results of Monte Carlo simulations are presented which demonstrate that a Kalman filter with a target-oriented process noise tracks coordinated turning targets with smaller root-mean-square errors than a Kalman filter with standard process noise. It was determined that if the target-oriented process noise is to be effective, the Kalman filter must generate an acceleration estimate and the target's maneuver should be confined to a plane. Kalman Filter, Target, Tracking, Target-Oriented Process Noise, Process Noise, Monte Carlo simulations.