A Hybrid Kalman/γH∞ Filter for Maneuvering Target Tracking

In air surveillance systems, Kalman filter and its derives have been largely used for real-time multiple target tracking, where minimization of calculation burdens is greatly desired. While non-maneuvering targets can be accurately tracked by using a Kalman filter with a constant velocity model, a maneuvring target might be miss-tracked since prediction errors increase significantly. The H_∞ filter is the robust filter, which does change its characteristics depending on the parameterγ. In order to track accurately not only non-maneuvering but also highly maneuvering target, we propose, in this paper, a variable γH_∞, filter with an acceleration term when a maneuver is detected.