Comparison of Auxiliary and Likelihood Particle Filters for State Estimation of Dynamical Systems

摘要

In this paper, algorithms of the state estimation of dynamical systems, using different types of particle filters, have been presented. Three Particle Filter methods have been used: Bootstrap Filter, Auxiliary Particle Filter and Likelihood Particle Filter. These methods have been applied to two nonlinear objects, with quadratic measurement functions. The results have been additionally compared with the outcome from Kalman filters. Based on the obtained results (5 different quality indices) the estimation methods have been evaluated.