Skip to content

Detail of publication

Citation

Šimandl, M. and Straka, O. : Sampling density design for particle filters . 13th IFAC Symposium on System Identification, p. 180-185, OMNIPRESS, Rotterdam, 2003.

Abstract

Particle filters for state estimation of discrete time dynamic stochastic systems are treated. The stress is laid on design of sampling pdf which is significant for quality of the particle filters. A new functional sampling density design based on utilization of transient and measurement pdf's is proposed. The functional approach compares two pdf's of a reference variable which are obtained by transformation of the transient and measurement pdf's, using Kullback J-divergence. The functional approach to sampling density function synthesis using transient and measurement density functions can be understood as improvement of the sampling density design of the auxiliary particle filter which uses only point estimate of the transient pdf. The higher quality of the functional particle filter with respect to the bootstrap and auxiliary particle filter is illustrated in a numerical example.

Detail of publication

Title: Sampling density design for particle filters
Author: Šimandl, M. ; Straka, O.
Language: English
Date of publication: 27 Aug 2003
Year: 2003
Type of publication: Papers in proceedings of reviewed conferences
Title of journal or book: 13th IFAC Symposium on System Identification
Page: 180 - 185
Publisher: OMNIPRESS
Address: Rotterdam
Date: 27 Aug 2003 - 29 Aug 2003
/ /

Keywords

state estimation, nonlinear systems, particle filters, sampling density

BibTeX

@INPROCEEDINGS{SimandlM_2003_Samplingdensity,
 author = {\v{S}imandl, M. and Straka, O.},
 title = {Sampling density design for particle filters},
 year = {2003},
 publisher = {OMNIPRESS},
 journal = {13th IFAC Symposium on System Identification},
 address = {Rotterdam},
 pages = {180-185},
 url = {http://www.kky.zcu.cz/en/publications/SimandlM_2003_Samplingdensity},
}