Robust Poisson multi-Bernoulli mixture filter using adaptive birth distributions for extended targets

Abstract

This paper presents a robust Poisson multi-Bernoulli mixture (PMBM) filter using adaptive birth distributions for the tracking of multiple extended targets. Firstly, in order to be immune to the uncertainty of target birth, we present a novel measurement-driven adaptive birth distribution that is robust to the arbitrariness of locations where new targets appear. Then, to enhance practicability of algorithm, the Beta distribution is employed to describe unknown detection probability augmented into the extended target state. On these bases, the detailed recursion and closed-form solutions to the proposed filter are derived by means of approximating the intensity of target birth and potential targets to Beta gamma Gaussian inverse Wishart (BGGIW) mixture form and density of existing Bernoulli component to a single BGGIW form. Moreover, the computational bottleneck caused by data associations is settled by truncating the PMBM filtering density using the Gibbs sampler of the Metropolis Hasting Markov Chain Monte Carlo (MCMC). Simulation results show the robustness and effectiveness of the proposed method.