Multiple target tracking using adaptive multiple maneuver model joint probabilistic data association
Yoshio Kosuge, Masamichi Kojima, Seiji Mano · Electronics and Communications in Japan (Part I Communications) · 2000
The M3JPDA (multiple maneuver model joint probabilistic data association) system is a multiple target tracking algorithm that can cope with maneuvering targets. The system includes a straight-line constant-velocity movement model as well as multiple accelerated movement models, to be used in parallel, where the models are utilized according to their reliability. It sometimes happens in M3JPDA, however, that when the targets move in straight lines with constant velocities, the multiple tracking performance is worse than in the conventional JPDA using the straight-line constant-velocity movement model. This paper proposes adaptive M3JPDA to remedy this point. In this method, the target acceleration is estimated and is smoothed. If the resulting value is large, the conventional M3JPDA is used. If the value is small, the reliability of each motion model is calculated and the choice of models is based on the difference between the reliabilities of the conventional M3JPDA accelerated motion models and the straight-line constant-velocity motion model. Using the proposed algorithm, the reliability of the straight-line constant-velocity motion model is improved when the target executes straight-line constant-velocity motion. As a result of evaluation by simulation, it is seen that the proposed algorithm is capable of tracking two crossing targets in straight-line constant-velocity motions, as well as two targets rapidly maneuvering with 7 g accelerations, in contrast to JPDA based on the straight-line constant-velocity motion model and M3JPDA. © 1999 Scripta Technica, Electron Comm Jpn Pt 1, 83(2): 61–73, 2000