Exploiting geometric translations in bearings-only TLS tracking

This paper studies the effects of local coordinate translations on the performance of the bearings-only total least squares (TLS) target tracking algorithm. The TLS estimator has been proposed to alleviate the severe bias problems associated with the traditional pseudolinear estimator. In the passive localization context the bias of the TLS estimator was shown to be sensitive to shifts of the origin of the local Cartesian coordinates. In this paper we extend this observation to bearings-only target tracking. After a brief overview of the LS and TLS tracking algorithms, we propose a simple and effective method for TLS bias reduction based on the dependence of the TLS bias on translations of the coordinate system origin. The improved TLS performance is illustrated with simulation examples.