Improvement of Doppler Velocity Log Measurements Using a Data-Driven Approach Based on Convolutional Neural Networks for Underwater Vehicle Navigation

hosseini, mostafa

EN FA

year 13, Issue 1 (Journal of Acoustical Society of Iran 2025) مجله انجمن علوم صوتی ایران (مهندسی صوتیات سابق) 2025, 13(1): 53-61 | Back to browse issues page

Mendeley

Zotero

RefWorks

hosseini M. Improvement of Doppler Velocity Log Measurements Using a Data-Driven Approach Based on Convolutional Neural Networks for Underwater Vehicle Navigation. مجله انجمن علوم صوتی ایران (مهندسی صوتیات سابق) 2025; 13 (1) : 6
URL: http://joasi.ir/article-1-312-en.html

Improvement of Doppler Velocity Log Measurements Using a Data-Driven Approach Based on Convolutional Neural Networks for Underwater Vehicle Navigation

Mostafa Hosseini ^*

Abstract: (32 Views)

Underwater vehicles (submersibles) typically have a wide range of commercial and military applications. Due to the limitations of satellite-based navigation systems underwater, the navigation and positioning of these vehicles pose significant challenges. One of the most effective and widely-used methods without spatial constraints for positioning such vessels is the integration of an Inertial Navigation System (INS) with a Doppler Velocity Log (DVL). In this integrated navigation system, the DVL calculates the vehicle's velocity vector, the accuracy of which is crucial for overall navigation precision and aids in estimating the navigation system's state variables. This study proposes a data-driven Convolutional Neural Network (CNN) approach for regressing the DVL-estimated velocity vector, which enhances estimation accuracy and can replace conventional model-based estimators such as the Least Squares (LS) method. The proposed method utilizes current DVL beam measurements and inertial sensor data (from accelerometers and gyroscopes) to estimate the vehicle's velocity. Simulations and experiments were conducted using real DVL data to validate the proposed method against the model-based approach. The results demonstrate that the proposed method achieves an improvement of over 60% in estimating the DVL velocity vector.

Article number: 6

Keywords: Doppler velocity log, Integrated navigation system, Convolutional neural network, Deep learning.

Full-Text [PDF 779 kb] (18 Downloads)

Type of Study: Research | Subject: Hydroacoustics
Received: 2024/11/10 | Accepted: 2025/06/20 | Published: 2025/12/21

Send email to the article author

Rights and permissions
	This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.