Adaptive Kernel Kalman Filter

Mengwei Sun; Michael E. Davies; Ian Proudler; James R. Hopgood

doi:10.1109/SSPD51364.2021.9541455

Adaptive Kernel Kalman Filter

Mengwei Sun, Michael E. Davies, Ian Proudler, James R. Hopgood

School of Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract / Description of output

This paper presents a novel model-based Bayesian filter called the adaptive kernel Kalman filter (AKKF). The proposed filter approximates the arbitrary probability distribution functions (PDFs) of hidden states as empirical kernel mean embeddings (KMEs) in reproducing kernel Hilbert spaces (RKHSs). Specifically, particles are generated and updated in the data space to capture the properties of the dynamical system model, while the corresponding kernel weight vector and matrix associated with the particles' feature mappings are predicted and updated in the RKHS based on the kernel Kalman rule (KKR). We illustrate and confirm the advantages of our approach through simulation, offering detailed comparison with the unscented Kalman filter (UKF), particle filter (PF) and Gaussian particle filter (GPF) algorithms.

Original language	English
Title of host publication	2021 Sensor Signal Processing for Defence Conference
Publisher	Institute of Electrical and Electronics Engineers
DOIs	https://doi.org/10.1109/SSPD51364.2021.9541455
Publication status	E-pub ahead of print - 23 Sept 2021
Event	International Conference in Sensor Signal Processing for Defence: : from Sensor to Decision - The Royal College of Physicians , Edinburgh, United Kingdom Duration: 14 Sept 2021 → 15 Sept 2021 https://www.sspd.eng.ed.ac.uk/

Conference

Conference	International Conference in Sensor Signal Processing for Defence:
Abbreviated title	SSPD2021
Country/Territory	United Kingdom
City	Edinburgh
Period	14/09/21 → 15/09/21
Internet address	https://www.sspd.eng.ed.ac.uk/

Access to Document

10.1109/SSPD51364.2021.9541455

1 Finished

Signal Processing in the Information Age
Davies, M., Hopgood, J., Hospedales, T., Mulgrew, B., Thompson, J., Tsaftaris, S. & Yaghoobi Vaighan, M.
EPSRC
1/07/18 → 31/03/24
Project: Research

3 Conference contribution
2 Article

Implementation of AKKF-based Multi-Sensor Fusion Methods in Stone Soup
Wright, J., Sun, M., Davies, M. E., Proudler, I. & Hopgood, J. R., 11 Oct 2024, (E-pub ahead of print) 2024 27th International Conference on Information Fusion (FUSION). Institute of Electrical and Electronics Engineers, 6 p.
Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Open Access
File
Adaptive Kernel Kalman Filter for Magnetic Anomaly Detection-based Metallic Target Tracking
Sun, M., Hodgskin-Brown, R., Davies, M. E., Proudler, I. & Hopgood, J. R., 22 Sept 2023, (E-pub ahead of print) 2023 Sensor Signal Processing for Defence Conference (SSPD) . Institute of Electrical and Electronics Engineers, 5 p.
Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Open Access
File
Implementation of Adaptive Kernel Kalman Filter in Stone Soup
Wright, J., Hopgood, J. R., Davies, M. E., Proudler, I. & Sun, M., 22 Sept 2023, (E-pub ahead of print) 2023 Sensor Signal Processing for Defence Conference (SSPD). Institute of Electrical and Electronics Engineers, 6 p.
Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Open Access
File

Cite this

@inproceedings{64e45339096c497dbf8eabec0a47b13b,

title = "Adaptive Kernel Kalman Filter",

abstract = "This paper presents a novel model-based Bayesian filter called the adaptive kernel Kalman filter (AKKF). The proposed filter approximates the arbitrary probability distribution functions (PDFs) of hidden states as empirical kernel mean embeddings (KMEs) in reproducing kernel Hilbert spaces (RKHSs). Specifically, particles are generated and updated in the data space to capture the properties of the dynamical system model, while the corresponding kernel weight vector and matrix associated with the particles' feature mappings are predicted and updated in the RKHS based on the kernel Kalman rule (KKR). We illustrate and confirm the advantages of our approach through simulation, offering detailed comparison with the unscented Kalman filter (UKF), particle filter (PF) and Gaussian particle filter (GPF) algorithms.",

author = "Mengwei Sun and Davies, {Michael E.} and Ian Proudler and Hopgood, {James R.}",

year = "2021",

month = sep,

day = "23",

doi = "10.1109/SSPD51364.2021.9541455",

language = "English",

booktitle = "2021 Sensor Signal Processing for Defence Conference",

publisher = "Institute of Electrical and Electronics Engineers",

address = "United States",

note = "International Conference in Sensor Signal Processing for Defence: : from Sensor to Decision, SSPD2021 ; Conference date: 14-09-2021 Through 15-09-2021",

url = "https://www.sspd.eng.ed.ac.uk/",

}

TY - GEN

T1 - Adaptive Kernel Kalman Filter

AU - Sun, Mengwei

AU - Davies, Michael E.

AU - Proudler, Ian

AU - Hopgood, James R.

PY - 2021/9/23

Y1 - 2021/9/23

N2 - This paper presents a novel model-based Bayesian filter called the adaptive kernel Kalman filter (AKKF). The proposed filter approximates the arbitrary probability distribution functions (PDFs) of hidden states as empirical kernel mean embeddings (KMEs) in reproducing kernel Hilbert spaces (RKHSs). Specifically, particles are generated and updated in the data space to capture the properties of the dynamical system model, while the corresponding kernel weight vector and matrix associated with the particles' feature mappings are predicted and updated in the RKHS based on the kernel Kalman rule (KKR). We illustrate and confirm the advantages of our approach through simulation, offering detailed comparison with the unscented Kalman filter (UKF), particle filter (PF) and Gaussian particle filter (GPF) algorithms.

AB - This paper presents a novel model-based Bayesian filter called the adaptive kernel Kalman filter (AKKF). The proposed filter approximates the arbitrary probability distribution functions (PDFs) of hidden states as empirical kernel mean embeddings (KMEs) in reproducing kernel Hilbert spaces (RKHSs). Specifically, particles are generated and updated in the data space to capture the properties of the dynamical system model, while the corresponding kernel weight vector and matrix associated with the particles' feature mappings are predicted and updated in the RKHS based on the kernel Kalman rule (KKR). We illustrate and confirm the advantages of our approach through simulation, offering detailed comparison with the unscented Kalman filter (UKF), particle filter (PF) and Gaussian particle filter (GPF) algorithms.

U2 - 10.1109/SSPD51364.2021.9541455

DO - 10.1109/SSPD51364.2021.9541455

M3 - Conference contribution

BT - 2021 Sensor Signal Processing for Defence Conference

PB - Institute of Electrical and Electronics Engineers

T2 - International Conference in Sensor Signal Processing for Defence:

Y2 - 14 September 2021 through 15 September 2021

ER -

Adaptive Kernel Kalman Filter

Abstract / Description of output

Conference

Access to Document

Fingerprint

Projects

Signal Processing in the Information Age

Research output

Implementation of AKKF-based Multi-Sensor Fusion Methods in Stone Soup

Adaptive Kernel Kalman Filter for Magnetic Anomaly Detection-based Metallic Target Tracking

Implementation of Adaptive Kernel Kalman Filter in Stone Soup

Cite this