Integrated Gaussian Processes for Tracking
In applications such as tracking and localisation, a dynamical model is typically specified for the modelling of an object's motion. An appealing alternative to the traditional parametric Markovian dynamical models is the Gaussian Process (GP). GPs can offer additional flexibility and rep...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Open Journal of Signal Processing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10839315/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1823859625399681024 |
|---|---|
| author | Fred Lydeard Bashar I. Ahmad Simon Godsill |
| author_facet | Fred Lydeard Bashar I. Ahmad Simon Godsill |
| author_sort | Fred Lydeard |
| collection | DOAJ |
| description | In applications such as tracking and localisation, a dynamical model is typically specified for the modelling of an object's motion. An appealing alternative to the traditional parametric Markovian dynamical models is the Gaussian Process (GP). GPs can offer additional flexibility and represent non-Markovian, long-term, dependencies in the target's kinematics. However, a standard GP with constant or zero mean is prone to oscillating around its mean and not sufficiently exploring the state space. In this paper, we consider extensions of the common GP framework such that a GP acts as the driving <italic>disturbance</italic> term that is integrated over time to produce a new Integrated GP (iGP) dynamical model. It potentially provides a more realistic modelling of agile objects' behaviour. We prove here that the introduced iGP model is, itself, a GP with a non-stationary kernel, which we derive fully in the case of the squared exponential GP kernel. Thus, the iGP is straightforward to implement, with the usual growth over time of the computational burden. We further show how to implement the model with fixed time complexity in a standard sequential Bayesian updating framework using Kalman filter-based computations, employing a sliding window Markovian approximation. Example results from real radar measurements and synthetic data are presented to demonstrate the ability of the proposed iGP modelling to facilitate more accurate tracking compared to conventional GP. |
| format | Article |
| id | doaj-art-90d601948f8741d6aac739a97ace67d0 |
| institution | Kabale University |
| issn | 2644-1322 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Open Journal of Signal Processing |
| spelling | doaj-art-90d601948f8741d6aac739a97ace67d02025-02-11T00:01:45ZengIEEEIEEE Open Journal of Signal Processing2644-13222025-01-0169910710.1109/OJSP.2025.352930810839315Integrated Gaussian Processes for TrackingFred Lydeard0https://orcid.org/0000-0002-5177-0203Bashar I. Ahmad1https://orcid.org/0000-0001-8974-6041Simon Godsill2https://orcid.org/0000-0001-9522-9681Department of Engineering, University of Cambridge, Cambridge, U.K.Department of Engineering, University of Cambridge, Cambridge, U.K.Department of Engineering, University of Cambridge, Cambridge, U.K.In applications such as tracking and localisation, a dynamical model is typically specified for the modelling of an object's motion. An appealing alternative to the traditional parametric Markovian dynamical models is the Gaussian Process (GP). GPs can offer additional flexibility and represent non-Markovian, long-term, dependencies in the target's kinematics. However, a standard GP with constant or zero mean is prone to oscillating around its mean and not sufficiently exploring the state space. In this paper, we consider extensions of the common GP framework such that a GP acts as the driving <italic>disturbance</italic> term that is integrated over time to produce a new Integrated GP (iGP) dynamical model. It potentially provides a more realistic modelling of agile objects' behaviour. We prove here that the introduced iGP model is, itself, a GP with a non-stationary kernel, which we derive fully in the case of the squared exponential GP kernel. Thus, the iGP is straightforward to implement, with the usual growth over time of the computational burden. We further show how to implement the model with fixed time complexity in a standard sequential Bayesian updating framework using Kalman filter-based computations, employing a sliding window Markovian approximation. Example results from real radar measurements and synthetic data are presented to demonstrate the ability of the proposed iGP modelling to facilitate more accurate tracking compared to conventional GP.https://ieeexplore.ieee.org/document/10839315/Integrated Gaussian processsquared exponential kerneldynamical modeltracking |
| spellingShingle | Fred Lydeard Bashar I. Ahmad Simon Godsill Integrated Gaussian Processes for Tracking IEEE Open Journal of Signal Processing Integrated Gaussian process squared exponential kernel dynamical model tracking |
| title | Integrated Gaussian Processes for Tracking |
| title_full | Integrated Gaussian Processes for Tracking |
| title_fullStr | Integrated Gaussian Processes for Tracking |
| title_full_unstemmed | Integrated Gaussian Processes for Tracking |
| title_short | Integrated Gaussian Processes for Tracking |
| title_sort | integrated gaussian processes for tracking |
| topic | Integrated Gaussian process squared exponential kernel dynamical model tracking |
| url | https://ieeexplore.ieee.org/document/10839315/ |
| work_keys_str_mv | AT fredlydeard integratedgaussianprocessesfortracking AT bashariahmad integratedgaussianprocessesfortracking AT simongodsill integratedgaussianprocessesfortracking |