Monte Carlo mean for non-Gaussian autonomous object tracking

L. Marata, J. Chuma, I. Ngebani, A. Yahya, O. L.A. López

Research output: Contribution to journalArticle

Abstract

Object tracking is highly applicable in emerging technologies and is normally done using measurements from sensors. Unfortunately, due to the presence of deleterious noise, measurements are inaccurate and different estimation methods have been developed. Most of them are mainly for Gaussian noise, leaving non-Gaussian noise scenarios unresolved. Also, while particle filters were introduced to address a more general noise scenario, they are mathematically complex especially when used in high dimensional systems. To circumvent these problems, we propose the Separate Monte Carlo Mean (SMC-MEAN) which is formulated on the Bayesian particle filtering framework. The proposed method is applied to an autonomous object tracking problem in both Gaussian and non-Gaussian scenarios. Results are compared to the Kalman filter and Maximum A Posteriori (MAP) in Exponential and Logistic distributed noise. The proposed method outperforms the other methods by an average of 17% yet maintaining low mathematical complexity.

Original languageEnglish
Pages (from-to)389-397
Number of pages9
JournalComputers and Electrical Engineering
Volume76
DOIs
Publication statusPublished - Jun 1 2019

Fingerprint

Kalman filters
Logistics
Sensors

All Science Journal Classification (ASJC) codes

  • Control and Systems Engineering
  • Computer Science(all)
  • Electrical and Electronic Engineering

Cite this

@article{3415637c602b46a999fcc1e3f9e6e0e5,
title = "Monte Carlo mean for non-Gaussian autonomous object tracking",
abstract = "Object tracking is highly applicable in emerging technologies and is normally done using measurements from sensors. Unfortunately, due to the presence of deleterious noise, measurements are inaccurate and different estimation methods have been developed. Most of them are mainly for Gaussian noise, leaving non-Gaussian noise scenarios unresolved. Also, while particle filters were introduced to address a more general noise scenario, they are mathematically complex especially when used in high dimensional systems. To circumvent these problems, we propose the Separate Monte Carlo Mean (SMC-MEAN) which is formulated on the Bayesian particle filtering framework. The proposed method is applied to an autonomous object tracking problem in both Gaussian and non-Gaussian scenarios. Results are compared to the Kalman filter and Maximum A Posteriori (MAP) in Exponential and Logistic distributed noise. The proposed method outperforms the other methods by an average of 17{\%} yet maintaining low mathematical complexity.",
author = "L. Marata and J. Chuma and I. Ngebani and A. Yahya and L{\'o}pez, {O. L.A.}",
year = "2019",
month = "6",
day = "1",
doi = "10.1016/j.compeleceng.2019.04.004",
language = "English",
volume = "76",
pages = "389--397",
journal = "Computers and Electrical Engineering",
issn = "0045-7906",
publisher = "Elsevier Limited",

}

Monte Carlo mean for non-Gaussian autonomous object tracking. / Marata, L.; Chuma, J.; Ngebani, I.; Yahya, A.; López, O. L.A.

In: Computers and Electrical Engineering, Vol. 76, 01.06.2019, p. 389-397.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Monte Carlo mean for non-Gaussian autonomous object tracking

AU - Marata, L.

AU - Chuma, J.

AU - Ngebani, I.

AU - Yahya, A.

AU - López, O. L.A.

PY - 2019/6/1

Y1 - 2019/6/1

N2 - Object tracking is highly applicable in emerging technologies and is normally done using measurements from sensors. Unfortunately, due to the presence of deleterious noise, measurements are inaccurate and different estimation methods have been developed. Most of them are mainly for Gaussian noise, leaving non-Gaussian noise scenarios unresolved. Also, while particle filters were introduced to address a more general noise scenario, they are mathematically complex especially when used in high dimensional systems. To circumvent these problems, we propose the Separate Monte Carlo Mean (SMC-MEAN) which is formulated on the Bayesian particle filtering framework. The proposed method is applied to an autonomous object tracking problem in both Gaussian and non-Gaussian scenarios. Results are compared to the Kalman filter and Maximum A Posteriori (MAP) in Exponential and Logistic distributed noise. The proposed method outperforms the other methods by an average of 17% yet maintaining low mathematical complexity.

AB - Object tracking is highly applicable in emerging technologies and is normally done using measurements from sensors. Unfortunately, due to the presence of deleterious noise, measurements are inaccurate and different estimation methods have been developed. Most of them are mainly for Gaussian noise, leaving non-Gaussian noise scenarios unresolved. Also, while particle filters were introduced to address a more general noise scenario, they are mathematically complex especially when used in high dimensional systems. To circumvent these problems, we propose the Separate Monte Carlo Mean (SMC-MEAN) which is formulated on the Bayesian particle filtering framework. The proposed method is applied to an autonomous object tracking problem in both Gaussian and non-Gaussian scenarios. Results are compared to the Kalman filter and Maximum A Posteriori (MAP) in Exponential and Logistic distributed noise. The proposed method outperforms the other methods by an average of 17% yet maintaining low mathematical complexity.

UR - http://www.scopus.com/inward/record.url?scp=85064495222&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85064495222&partnerID=8YFLogxK

U2 - 10.1016/j.compeleceng.2019.04.004

DO - 10.1016/j.compeleceng.2019.04.004

M3 - Article

AN - SCOPUS:85064495222

VL - 76

SP - 389

EP - 397

JO - Computers and Electrical Engineering

JF - Computers and Electrical Engineering

SN - 0045-7906

ER -