Bearings Tracking of Targets with Unobservable States

Multi-target correlator-trackers often require state information for the correlation task, even before an estimate of position and velocity with error ellipse is available. Not only must one deal with the indeterminacy present in a single leg of bearing data, but also with those cases which, despite sensor maneuvers or reports from multiple sensors, remain effectively unobservable because of noise or unfavorable geometry. In this paper a new suboptimal tracking algorithm suitable for cases in which the state is not fully observable will be discussed, and practical aspects of its operation in a data association environment will be pointed out.