A Computationally Efficient and Robust Filtering Framework for Underwater Multi-sensor Navigation

doi:10.1109/tim.2026.3736119

A Computationally Efficient and Robust Filtering Framework for Underwater Multi-sensor Navigation

2026-09-22

0

PRE

AI

DOI：10.1109/tim.2026.3736119

Original

Original request for help

Save

Abstract

En 中文

Existing high-accuracy robust filtering algorithms, while precise, are computationally intensive and require numerous iterations, rendering them impractical for autonomous underwater vehicle (AUV) platforms where real-time processing and limited onboard resources are critical. To address this, a novel computationally efficient and robust filtering framework is proposed, significantly reducing iteration and computational demands while maintaining strong resilience to outlier-contaminated measurements. Theoretical analyses on stability, convergence, and computational complexity are presented to validate the framework. Experimental results demonstrate that the proposed approach enables high-accuracy and robust state estimation on AUVs with drastically reduced computational load and iteration count, outperforming existing robust filtering algorithms in practical AUV scenarios.

Keywords:

Underwater Multi-sensor Navigation

Federated Filter

Robust Filter

Information Fusion

State Estimation

A Computationally Efficient and Robust Filtering Framework for Underwater Multi-sensor Navigation

Abstract

Journal

Organization

Cited Papers