Two-stage Kalman estimator using advanced circular prediction for maneuvering target tracking

Abstract

Maneuvering targets are difficult to track for tihe Kalman filter since the target model of tracking filter might not fit the real barget trajectory and the stat,istical characteristics of t,he target maneuver are unknown in advance. In order to track such a heavy maiieuvering target, the est,iination of the target turndirection is necessary. T h e t,wo-stage estimator using advanced circular prediction which considers the ta.rget t,urn-direction is proposed for maneuvering target, tracking. Sirnulat,ion results are given for a comparison of the perfornimces of our proposed scheme with t>hat of c,onvent,ional t r x k i n g filkrs.

<title>Survey of maneuvering target tracking: dynamic models</title>

“…Note that if the sojourn time is nonrandom or has an exponential distribution, the input sequence (U(tk)) is in fact a Markov chain -the semi-Markov formulation is not needed. Moreover, they proposed the following model of the acceleration a(t) as a combination of the above jump-mean model and the Singer model: a(t) = -fiv(t) + u(t) + (42) where à(t) is the Singer acceleration of (25), v is the velocity, and /3 is a drag coefficient. The corresponding continuous-time state-space representation is, for x = [position,velocity,acceleration}',…”

Survey of maneuvering targettracking . part I: dynamic models

“…Assume that each target position measurements are points on the circle; replace the chord between any two consecutive measurement points with the straight line segment connecting them; the center can then be determined from the (average) intersection of the perpendicular bisectors of two or more such straight line segments. An essentially the same procedure was used in [81] for estimating the center. Note that using the center estimates injects additional nonlinearities into the system, which are not accounted for in the above linear model.…”

Intentional motion on-line learning and prediction

“…Approaches based on the kinematic or dynamic properties of moving objects are often based on the Kalman Filter and Extended Kalman Filter's prediction step [9,21,24]. There are, however other approaches that also use dynamics for prediction: Chien and Koivo [11] proposed the use of a recursive autoregressive time series model whose parameters are estimated using the least mean squared error method.…”