An Exponentially-Weighted Probabilistic Data Association Filter for Maneuvering Targets

An algorithm is presented for tracking multiple targets which are rapidly maneuvering in an environment in which false measurements may occur. The real-time algorithm combines a fading memory probabilistic data association filter and a recursive least-squares estimation procedure to estimate the state and the acceleration of the maneuvering target. Identification of the existence of a maneuver is not required.