Target tracking via a dynamic circular filter/linear /spl alpha/-/spl beta/ filters in 2D

This paper focuses on the development of a technique for tracking maneuvering target in two dimensions. Since the track of a maneuvering target in two dimensions can be approximated by an arc of a circle, for at least some finite distance, it seems obvious to take into account its movement angularly as well as linearly. The proposed algorithm (the /spl delta/-/spl epsiv/ filter) is similar to the /spl alpha/-/spl beta/ filter, except the filter gains, /spl delta/ and /spl epsiv/, instead weigh the angular displacement and angular velocity, respectively. To account for straight line target trajectories, a linear /spl alpha/-/spl beta/ filter is used in parallel, and it is at this point a weighted combination of the two predictions is used to determine final future predicted position.