Monte Carlo filter particle filter

Masaya Murata; Hidehisa Nagano; Kunio Kashino

doi:10.1109/ecc.2015.7330967

Monte Carlo filter particle filter

Masaya Murata
, Hidehisa Nagano
, Kunio Kashino

Not published at Aalto University

NTT Corporation
National Institute of Informatics

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

Abstract

We propose a new realization method of the sequential importance sampling (SIS) algorithm to derive a new particle filter. The new filter constructs the importance distribution by the Monte Carlo filter (MCF) using sub-particles, therefore, its non-Gaussianity nature can be adequately considered while the other type of particle filter such as unscented Kalman filter particle filter (UKF-PF) assumes a Gaussianity on the importance distribution. Since the state estimation accuracy of the SIS algorithm theoretically improves as the estimated importance distribution becomes closer to the true posterior probability density function of state, the new filter is expected to outperform the existing, state-of-the-art particle filters. We call the new filter Monte Carlo filter particle filter (MCF-PF) and confirm its effectiveness through the numerical simulations.

Original language	English
Title of host publication	2015 European Control Conference (ECC)
Publisher	IEEE
Number of pages	6
ISBN (Electronic)	9783952426937
DOIs	https://doi.org/10.1109/ecc.2015.7330967
Publication status	Published - 2015
MoE publication type	A4 Conference publication
Event	European Control Conference - Linz, Austria Duration: 15 Jul 2015 → 17 Jul 2015 Conference number: 14

Conference

Conference	European Control Conference
Abbreviated title	ECC
Country/Territory	Austria
City	Linz
Period	15/07/2015 → 17/07/2015

Access to Document

10.1109/ecc.2015.7330967

Cite this

@inproceedings{165947fb6ad54eac817c8ebe8f50e854,

title = "Monte Carlo filter particle filter",

abstract = "We propose a new realization method of the sequential importance sampling (SIS) algorithm to derive a new particle filter. The new filter constructs the importance distribution by the Monte Carlo filter (MCF) using sub-particles, therefore, its non-Gaussianity nature can be adequately considered while the other type of particle filter such as unscented Kalman filter particle filter (UKF-PF) assumes a Gaussianity on the importance distribution. Since the state estimation accuracy of the SIS algorithm theoretically improves as the estimated importance distribution becomes closer to the true posterior probability density function of state, the new filter is expected to outperform the existing, state-of-the-art particle filters. We call the new filter Monte Carlo filter particle filter (MCF-PF) and confirm its effectiveness through the numerical simulations.",

author = "Masaya Murata and Hidehisa Nagano and Kunio Kashino",

year = "2015",

doi = "10.1109/ecc.2015.7330967",

language = "English",

booktitle = "2015 European Control Conference (ECC)",

publisher = "IEEE",

address = "United States",

note = "European Control Conference, ECC ; Conference date: 15-07-2015 Through 17-07-2015",

}

TY - GEN

T1 - Monte Carlo filter particle filter

AU - Murata, Masaya

AU - Nagano, Hidehisa

AU - Kashino, Kunio

N1 - Conference code: 14

PY - 2015

Y1 - 2015

N2 - We propose a new realization method of the sequential importance sampling (SIS) algorithm to derive a new particle filter. The new filter constructs the importance distribution by the Monte Carlo filter (MCF) using sub-particles, therefore, its non-Gaussianity nature can be adequately considered while the other type of particle filter such as unscented Kalman filter particle filter (UKF-PF) assumes a Gaussianity on the importance distribution. Since the state estimation accuracy of the SIS algorithm theoretically improves as the estimated importance distribution becomes closer to the true posterior probability density function of state, the new filter is expected to outperform the existing, state-of-the-art particle filters. We call the new filter Monte Carlo filter particle filter (MCF-PF) and confirm its effectiveness through the numerical simulations.

AB - We propose a new realization method of the sequential importance sampling (SIS) algorithm to derive a new particle filter. The new filter constructs the importance distribution by the Monte Carlo filter (MCF) using sub-particles, therefore, its non-Gaussianity nature can be adequately considered while the other type of particle filter such as unscented Kalman filter particle filter (UKF-PF) assumes a Gaussianity on the importance distribution. Since the state estimation accuracy of the SIS algorithm theoretically improves as the estimated importance distribution becomes closer to the true posterior probability density function of state, the new filter is expected to outperform the existing, state-of-the-art particle filters. We call the new filter Monte Carlo filter particle filter (MCF-PF) and confirm its effectiveness through the numerical simulations.

U2 - 10.1109/ecc.2015.7330967

DO - 10.1109/ecc.2015.7330967

M3 - Conference article in proceedings

BT - 2015 European Control Conference (ECC)

PB - IEEE

T2 - European Control Conference

Y2 - 15 July 2015 through 17 July 2015

ER -

Monte Carlo filter particle filter

Abstract

Conference

Access to Document

Fingerprint

Cite this