Concise track characterization of maneuvering targets
Stephen Paul Linder, Richard Quintin, Matthew Ryan · AIAA Guidance, Navigation, and Control Conference and Exhibit · 2001
The identification and characterization of targets is difficult using conventional Kalman filter-based trackers, because these trackers are not able to provide an accurate and succinct list of maneuvers for use in track and target identification. We propose a new approach, representing target tracks as a sequence of circular and linear splines which directly correspond to the sequence of coordinated turns and constant velocity motion that a target may perform. The splines provide a locally optimal estimate of the track by (1) minimizing the total least- squares error of the data fit, and (2) minimizing the dis- continuity in position and heading between successive splines. In addition, a parsimonious representation is ob- tained by dynamically merging redundant splines using a recursive algorithm. Our simulations studies show a large improvement in estimation accuracy, especially with re- spect to turn rate and heading. Our new approach shows great promise as a technique for ascertaining concise track characterization of maneuvering targets.