Decentralized Gaussian Process With Noisy Kernel Method to Construct an Underwater Communication Map With Uncertain Measurements
IEEE Robotics and Automation Letters · 2026-12 · 11 (12) · 14796-14803
DOI: 10.1109/lra.2026.3746067 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Robotics and Automation Letters, 2026-12. Verfasst von Yifei Gao und 2 weiteren Autor:innen. Die Arbeit nennt 21 Literaturangaben. Gefördert von: Seale Coastal Zone Observatory and in part by the Office of Naval Research.
Abstract (Originalsprache)
We present a decentralized sparse Gaussian process classification framework for constructing underwater acoustic communication maps using teams of autonomous underwater vehicles (AUVs). Although acoustic communication is known to be challenging in the underwater environment, it is the primary mode of communication available to AUVs while underwater. Our work provides an approach for AUVs to collaboratively characterize the performance of underwater acoustic communication in situ, which allows them to exploit favorable communication environments when they arise, and to compensate for poor communication environments when necessary. We adapt an expected noisy radial basis function(RBF) kernel which marginalizes input noise in closed-form and use it to derive an inducing point selection strategy. Selected inducing points are shared among AUVs, resulting in a bandwidth-efficient data sharing policy that is suitable for the bandwidth constraints imposed by acoustic communication with minimal reduction in map accuracy. We perform experiments on data collected in the Chesapeake Bay by a team of AUVs and show improved prediction performance over standard sparse Gaussian process methods.
Förderung
- Seale Coastal Zone Observatory and in part by the Office of Naval Research
Zitieren
APA
Gao, Y., Morency, C., & Stilwell, D. J. (2026). Decentralized Gaussian Process With Noisy Kernel Method to Construct an Underwater Communication Map With Uncertain Measurements. IEEE Robotics and Automation Letters, 11(12), 14796-14803. https://doi.org/10.1109/lra.2026.3746067
Vancouver
Gao Y, Morency C, Stilwell DJ. Decentralized Gaussian Process With Noisy Kernel Method to Construct an Underwater Communication Map With Uncertain Measurements. IEEE Robotics and Automation Letters. 2026;11(12):14796-14803. doi:10.1109/lra.2026.3746067
BibTeX
@article{Gao2026,
author = {Gao, Yifei and Morency, Christopher and Stilwell, Daniel J.},
title = {Decentralized Gaussian Process With Noisy Kernel Method to Construct an Underwater Communication Map With Uncertain Measurements},
journal = {IEEE Robotics and Automation Letters},
year = {2026},
volume = {11},
number = {12},
pages = {14796--14803},
doi = {10.1109/lra.2026.3746067}
}
EUVIMED