Single And Multiple Target Tracking

Author : Mohamed Nabil Abdelghaffar Eltoukhy
Publisher :
ISBN 13 :
Total Pages : 0 pages
Book Rating : 4.:/5 (139 download)

DOWNLOAD NOW!

Download or read book Algorithms for Tracking Single Maneuvering and Multiple Closely-Spaced Targets written by Mohamed Nabil Abdelghaffar Eltoukhy and published by . This book was released on 2021 with total page 0 pages. Available in PDF, EPUB and Kindle. Book excerpt: Target tracking is crucial in monitoring and controlling air traffic in civilian and military applications. Target tracking is a process of estimating the current position and predict the future position of one or more targets using the measurements received by a radar system. One of the major challenges in tracking a single target is when it performs a maneuver and the angle of maneuver is not known. The interacting multiple model (IMM) algorithm is the most commonly-used algorithm for tracking a maneuvering target with an a priori knowledge of the target turn rate, since it provides a very good tracking performance with moderate complexity. However, the tracking performance of such an algorithm deteriorates or may even fail when the target performs a maneuver with a turn rate larger than that assumed in the design of the algorithm. A few methods have been reported to overcome this limitation of an assumed turn rate by actually estimating it adaptively. Two of such algorithms use nonlinear filters that leads to a large complexity, and one of them uses linear filters and models providing good tracking performance, but only for mild maneuvers. For tracking multiple targets, several algorithms have been proposed, among which the joint probability data association (JPDA) algorithm is considered to be the best algorithm, since it provides good tracking performance when the targets are widely spaced. However, the tracking performance of this algorithm deteriorates, and coalescence of the tracks may occur, when the targets are closely spaced. Some efforts have been made to overcome the problem of tracking closely spaced targets by ignoring the target identity, but at the expense of very large complexity. The work of this thesis is carried out in two parts. In the first part, two algorithms within the IMM framework are proposed to track a single maneuvering target, when the target turn rate is not known a priori. In both the algorithms, the turn rate is dynamically estimated using noisy measurements. In the first algorithm, the turn rate at each time instant k is estimated based on the target speed and the radius of the circle formed by the measurement at that instant and the two previous consecutive noisy measurements, (k-1) and (k-2). The segment of this circle covered by these three noisy measurements is used to model the true track of the target at the instant time k. In the second algorithm, the accuracy of the turn rate estimated in the first algorithm is improved using the information on the level of the measurement noise. In the second part of the thesis, a systematic study on the impact of the spacing between the targets as well as when the targets make abrupt turns with sharp angles on the tracking performance of the JPDA algorithm is conducted. Then, a new algorithm for tracking multiple targets based on the spatial distribution of the measurements for determining the weights for measurement-target association is proposed within the JPDA framework. The proposed algorithm for multiple target tracking is designed to deal with the problems of closely spaced targets and their abrupt sharp turns more effectively. Effectiveness and superiority of the algorithms proposed for tracking single and multiple targets are demonstrated through extensive experiments with a wide variety of different scenarios for target motions.