Extended Object Tracking Using VB-MM-GIWIW-PMBM Filter

The Poisson multi-Bernoulli mixture (PMBM) filter algorithm incorporates the interactive multi-model (IMM) concept to address the issue of multiple extended objects tracking with random time-varying motion under multiple models. A gamma inverse Wishart inverse Wishart Poisson multi-Bernoulli mixture filter (VB-MM-GIWIW-PMBM) algorithm is proposed based on the variational Bayesian model, considering the mismatch between the angular velocity in the algorithm model set and the actual angular velocity of the target during maneuvering as well as the unknown covariance of measurement noise and motion noise. Simulation results show that the proposed method achieves tracking performance in scenarios of multiple extended object maneuvers in scenarios of imprecise processes, measurement noise covariance, and unknown turning angle velocity.