A Nonlinear Filtering Algorithm For Multi-Models

In this paper we study the problem of tracking maneuvering targets in the framework of nonlinear filtering where the signal has a continuous and a discrete part. We obtain the explicit representation for the filter density and introduce an approximation for the purposes of computation. We propose two different numerical methods for computation of the approximation. We use a simulation of a nonlinear problem to compare the results of these new algorithms to the Interactive Multiple Model (IMM) algorithm [2] that uses the extended Kalman filter. We observe that the nonlinear filtering algorithms drastically outperform the IMM algorithm