A sample size adaptation scheme for particle filter

Author

Duan, Zhuohua

Author_Institution

Sch. of Comput. Sci., Shaoguan Univ., Shaoguan, China

fYear

2012

fDate

23-25 May 2012

Firstpage

3012

Lastpage

3016

Abstract

Particle filter is a Monte Carlo method to monitor dynamic systems, which non-parametrically approximates probabilistic distribution using weighted samples. Particle filters have been widely used in various fields such as robotics, visual tracking, etc. A key issue for fast implementation of particle filter is how to determine the sample size (particle number) according the sample based distribution. The paper presents a sample size adaptation scheme for particle filters. The key idea is to adjust sample size according to the distance of two sample-based distributions with different sample scale. The method is testified on nonlinear system estimation problem.

Keywords

Monte Carlo methods; estimation theory; nonlinear systems; particle filtering (numerical methods); sampling methods; statistical distributions; Carlo method; dynamic systems monitor; nonlinear system estimation problem; nonparametric probabilistic distribution approximation; particle filter; sample size adaptation scheme; sample-based distributions; Accuracy; Monte Carlo methods; Noise; Particle filters; Presence network agents; State estimation; Target tracking; Adaptive; Particle Filter; Sample Size;

fLanguage

English

Publisher

ieee

Conference_Titel

Control and Decision Conference (CCDC), 2012 24th Chinese

Conference_Location

Taiyuan

Print_ISBN

978-1-4577-2073-4

Type

conf

DOI

10.1109/CCDC.2012.6244474

Filename

6244474