Particle filter tracking for banana and contact lens problems

In this paper we present an approach for tracking with high-bandwidth radar in long-range scenarios. We consider both two- and three-dimensional measurements in polar and r-u-v, respectively. We show that in these scenarios the extended Kalman filter is not desirable, because it suffers from major consistency problems, and most types of particle filters suffer from a loss of diversity among particles after resampling. This leads to sample impoverishment and divergence of the filter. In the scenarios studied, this loss of diversity can be attributed to the very low process noise. However, a regularized particle filter is shown to avoid this diversity problem while producing consistent results. The regularization is accomplished using a generalized version of the Epanechnikov kernel.