Theoretical study on multi-target tracking algorithm to get accurate target state estimation

Abstract

JPDAF is a method of updating target's state estimation by using posterior probability that measurements are originated from existing target in multi-target tracking. In this thesis, we propose a multi-target tracking algorithm for falling cluster bombs separated from a mother bomb based on JPDAS method which is obtained by applying fixed-interval smoothing technique to JPDAF. The performance of JPDAF and JPDAS multi-target tracking algorithm is compared by observing the average of the difference between targets' state estimations obtained from 100 independent executions of two algorithms and targets' true states. Based on this, results of simulations for a radar tracking problem that show the proposed JPDAS algorithm has better tracking performance than JPDAF is presented.