Multiple Model Truncated Particle Filter for maneuvering target tracking

A Multiple Model Truncated Particle Filter (MMTPF) algorithm is proposed for maneuvering target tracking. This algorithm merges the particles predicted by the multiple models into single particle set, calculates and normalizes the particle weights in that set at first. After the high weight particles established by the truncated threshold, this algorithm utilizes the QMC method and uniform kernel to construct the posterior probability density functions. The traditional interacting multiple model particle filter (IMMPF) is the integration of multiple single-model particle filters, while the new proposed algorithm is the particle filter with the fusion of multi-model. Results in maneuvering tracking show that the MMTPF algorithm achieves better estimation quality and computational efficiency than the IMMPF algorithm.