Particle Filtering within a Set-Membership Approach to State Estimation

This paper introduces a new algorithm where particle filtering techniques and set-membership theory are blended together in one only framework. The idea is to build a recursive filter where, at every step, an approximation of the probability density of the states given the latest observations is provided together with the set of all the possible states consistent with the process and observation models. The results obtained confirm that the advantages furnished by particle filtering and set-membership techniques add up together to obtain more accurate estimates