A Kalman Estimation Based Rao-Blackwellized Particle Filtering for Radar Tracking

Abstract

The Rao-Blackwellized particle filtering (RBPF) offers a general tracking framework with linear/nonlinear state space models, which outperforms the standard particle filtering in nonlinear and non-Gaussian tracking scenarios. Unfortunately, in conventional radar systems, the observations contain no information about the linear part of target state. In these cases, the RBPF algorithm fails to catch the real trajectories, because we cannot obtain the enough information to correctly update the linear part of target state in the tracking procedure. To overcome such an issue, this paper proposes a Kalman estimation-based BRPF (KE-BRPF) algorithm. In KE-RBPF, the correlation between linear and nonlinear parts of target state is investigated. Benefitting from such investigation, we derive a new set of formulea to present the correlation in terms of means and variances. By utilizing these formulas, our KE-RBPF algorithm correctly tracks the linear part of target state based on the nonlinear one. Finally, the simulation results verify that, our KE-RBPF performs better than other state-of-the-art tracking methods in nonlinear and non-Gaussian radar tracking scenarios, with at least 18% reduction in terms of the means and central tendency of error of tracking root-mean-square-error.