Authors: Seyyed Soheil Sadat Hosseini; Mohsin M. Jamali; Jaakko Astola; Peter V. Gorsevski
Addresses: Department of Electrical Engineering and Computer Science, The University of Toledo, Toledo, Ohio, USA ' Department of Electrical Engineering and Computer Science, The University of Toledo, Toledo, Ohio, USA ' Department of Signal Processing, Tampere University of Technology, Tampere, Finland ' Geospatial Science School of Earth & Environment, Bowling Green State University, Bowling Green, Ohio, USA
Abstract: Particle filters (PFs) have been used for the nonlinear estimation for a number of years. However, they suffer from the impoverishment phenomenon. It is brought by resampling which intends to prevent particle degradation, and therefore becomes the inherent weakness of this technique. To solve the problem of sample impoverishment and to improve the performance of the standard particle filter we propose a modification to this method by adding a sampling mechanism inspired by optimisation techniques, namely, the pattern search, particle swarm optimisation, differential evolution and Nelder-Mead algorithms. In the proposed methods, the true state of the target can be better expressed by the optimised particle set and the number of meaningful particles can be grown significantly. The efficiency of the proposed particle filters is supported by a truck-trailer problem. Simulations show that the hybridised particle filter with Nelder-Mead search is better than other optimisation approaches in terms of particle diversity.
Keywords: target tracking; particle filters; particle swarm optimisation; PSO; Nelder-Mead; pattern search; differential evolution; sample impoverishment; truck-trailer problem; simulation.
International Journal of Mathematical Modelling and Numerical Optimisation, 2016 Vol.7 No.2, pp.212 - 229
Received: 09 Jul 2015
Accepted: 14 Dec 2015
Published online: 18 Jun 2016 *