A Box Regularized Particle Filter for state estimation with severely ambiguous and non-linear measurements

Nicolas Jonathan Adrien Merlinge, Karim Dahia, Helene Piet-Lahanier, James Brusey, Nadjim Horri

Research output: Contribution to journalArticle

Abstract

The first stage in any control system is to be able to accurately estimate the system’s state. However, some types of measurements are ambiguous (non-injective) in terms of state. Existing algorithms for such problems, such as Monte Carlo methods, are computationally expensive or not robust to such ambiguity. We propose the Box Regularized Particle Filter (BRPF) to resolve these problems. Based on previous works on box particle filters, we present a more generic and accurate formulation of the algorithm, with two innovations: a generalized box resampling step and a kernel smoothing method, which is shown to be optimal in terms of Mean Integrated Square Error. Monte Carlo simulations demonstrate the efficiency of BRPF on a severely ambiguous and non-linear estimation problem, that of Terrain Aided Navigation. BRPF is compared to the Sequential Importance Resampling Particle Filter (SIR-PF), Monte Carlo Markov Chain (MCMC), and the original Box Particle Filter (BPF). The algorithm outperforms existing methods in terms of Root Mean Square Error (e.g., improvement up to 42% in geographical position estimation with respect to the BPF) for a large initial uncertainty. The BRPF reduces the computational load by 73%and 90% for SIR-PF and MCMC, respectively, with similar RMSE values. This work offers an accurate (in terms of RMSE) and robust (in terms of divergence rate) way to tackle state estimation from ambiguous measurements while requiring a significantly lower computational load than classic Monte Carlo and particle filtering methods.
Original languageEnglish
Pages (from-to)102-110
Number of pages9
JournalAutomatica
Volume104
Early online date12 Mar 2019
DOIs
Publication statusPublished - Jun 2019

Fingerprint

State estimation
Markov processes
Mean square error
Navigation
Monte Carlo methods
Innovation
Control systems

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Electrical and Electronic Engineering

Cite this

A Box Regularized Particle Filter for state estimation with severely ambiguous and non-linear measurements. / Merlinge, Nicolas Jonathan Adrien; Dahia, Karim; Piet-Lahanier, Helene; Brusey, James; Horri, Nadjim.

In: Automatica, Vol. 104, 06.2019, p. 102-110.

Research output: Contribution to journalArticle

Merlinge, Nicolas Jonathan Adrien ; Dahia, Karim ; Piet-Lahanier, Helene ; Brusey, James ; Horri, Nadjim. / A Box Regularized Particle Filter for state estimation with severely ambiguous and non-linear measurements. In: Automatica. 2019 ; Vol. 104. pp. 102-110.
@article{a31d8915b9ed4b61a1ef03cc8efba14b,
title = "A Box Regularized Particle Filter for state estimation with severely ambiguous and non-linear measurements",
abstract = "The first stage in any control system is to be able to accurately estimate the system’s state. However, some types of measurements are ambiguous (non-injective) in terms of state. Existing algorithms for such problems, such as Monte Carlo methods, are computationally expensive or not robust to such ambiguity. We propose the Box Regularized Particle Filter (BRPF) to resolve these problems. Based on previous works on box particle filters, we present a more generic and accurate formulation of the algorithm, with two innovations: a generalized box resampling step and a kernel smoothing method, which is shown to be optimal in terms of Mean Integrated Square Error. Monte Carlo simulations demonstrate the efficiency of BRPF on a severely ambiguous and non-linear estimation problem, that of Terrain Aided Navigation. BRPF is compared to the Sequential Importance Resampling Particle Filter (SIR-PF), Monte Carlo Markov Chain (MCMC), and the original Box Particle Filter (BPF). The algorithm outperforms existing methods in terms of Root Mean Square Error (e.g., improvement up to 42{\%} in geographical position estimation with respect to the BPF) for a large initial uncertainty. The BRPF reduces the computational load by 73{\%}and 90{\%} for SIR-PF and MCMC, respectively, with similar RMSE values. This work offers an accurate (in terms of RMSE) and robust (in terms of divergence rate) way to tackle state estimation from ambiguous measurements while requiring a significantly lower computational load than classic Monte Carlo and particle filtering methods.",
author = "Merlinge, {Nicolas Jonathan Adrien} and Karim Dahia and Helene Piet-Lahanier and James Brusey and Nadjim Horri",
year = "2019",
month = "6",
doi = "10.1016/j.automatica.2019.02.033",
language = "English",
volume = "104",
pages = "102--110",
journal = "Automatica",
issn = "0005-1098",
publisher = "Elsevier",

}

TY - JOUR

T1 - A Box Regularized Particle Filter for state estimation with severely ambiguous and non-linear measurements

AU - Merlinge, Nicolas Jonathan Adrien

AU - Dahia, Karim

AU - Piet-Lahanier, Helene

AU - Brusey, James

AU - Horri, Nadjim

PY - 2019/6

Y1 - 2019/6

N2 - The first stage in any control system is to be able to accurately estimate the system’s state. However, some types of measurements are ambiguous (non-injective) in terms of state. Existing algorithms for such problems, such as Monte Carlo methods, are computationally expensive or not robust to such ambiguity. We propose the Box Regularized Particle Filter (BRPF) to resolve these problems. Based on previous works on box particle filters, we present a more generic and accurate formulation of the algorithm, with two innovations: a generalized box resampling step and a kernel smoothing method, which is shown to be optimal in terms of Mean Integrated Square Error. Monte Carlo simulations demonstrate the efficiency of BRPF on a severely ambiguous and non-linear estimation problem, that of Terrain Aided Navigation. BRPF is compared to the Sequential Importance Resampling Particle Filter (SIR-PF), Monte Carlo Markov Chain (MCMC), and the original Box Particle Filter (BPF). The algorithm outperforms existing methods in terms of Root Mean Square Error (e.g., improvement up to 42% in geographical position estimation with respect to the BPF) for a large initial uncertainty. The BRPF reduces the computational load by 73%and 90% for SIR-PF and MCMC, respectively, with similar RMSE values. This work offers an accurate (in terms of RMSE) and robust (in terms of divergence rate) way to tackle state estimation from ambiguous measurements while requiring a significantly lower computational load than classic Monte Carlo and particle filtering methods.

AB - The first stage in any control system is to be able to accurately estimate the system’s state. However, some types of measurements are ambiguous (non-injective) in terms of state. Existing algorithms for such problems, such as Monte Carlo methods, are computationally expensive or not robust to such ambiguity. We propose the Box Regularized Particle Filter (BRPF) to resolve these problems. Based on previous works on box particle filters, we present a more generic and accurate formulation of the algorithm, with two innovations: a generalized box resampling step and a kernel smoothing method, which is shown to be optimal in terms of Mean Integrated Square Error. Monte Carlo simulations demonstrate the efficiency of BRPF on a severely ambiguous and non-linear estimation problem, that of Terrain Aided Navigation. BRPF is compared to the Sequential Importance Resampling Particle Filter (SIR-PF), Monte Carlo Markov Chain (MCMC), and the original Box Particle Filter (BPF). The algorithm outperforms existing methods in terms of Root Mean Square Error (e.g., improvement up to 42% in geographical position estimation with respect to the BPF) for a large initial uncertainty. The BRPF reduces the computational load by 73%and 90% for SIR-PF and MCMC, respectively, with similar RMSE values. This work offers an accurate (in terms of RMSE) and robust (in terms of divergence rate) way to tackle state estimation from ambiguous measurements while requiring a significantly lower computational load than classic Monte Carlo and particle filtering methods.

UR - http://www.scopus.com/inward/record.url?scp=85062634507&partnerID=8YFLogxK

U2 - 10.1016/j.automatica.2019.02.033

DO - 10.1016/j.automatica.2019.02.033

M3 - Article

VL - 104

SP - 102

EP - 110

JO - Automatica

JF - Automatica

SN - 0005-1098

ER -