Spectrally Normalized Memory Neuron Network Based Navigation for Autonomous Underwater Vehicles in DVL-Denied Environment

Makam, R and Pramuk, MP and Thomas, S and Sundaram, S (2024) Spectrally Normalized Memory Neuron Network Based Navigation for Autonomous Underwater Vehicles in DVL-Denied Environment. In: OCEANS 2024 - Singapore, OCEANS 2024, 15 April 2024 through 18 April 2024, Singapore.

PDF Oce_Con_Rec_(IEEE)_2024.pdf - Published Version Restricted to Registered users only Download (886kB) | Request a copy

Abstract

In this paper, we address the challenge of velocity estimation for Autonomous Underwater Vehicles (AUV) navi-gating in complex underwater environments without access to Doppler Velocity Log (DVL) information. The proposed solution adopts a learning-based approach taking inputs such as Inertial Measurement Unit (IMU) data, available DVL information, and past predicted velocities. It employs a novel spectrally normalized Memory Neuron network (SNMNN) to predict AUV velocity, ensuring stable and reliable performance through spectral nor-malization. The model is trained using the SNAPIR AUV dataset, incorporating IMU and DVL beams. The estimated velocity from the SNMNN is compared with the actual DVL velocity. The proposed method outperforms existing DVL-denied algorithms in terms of velocity estimation. It achieves lower root mean square error (RMSE) and higher variance accounted for (VAF) values. The results indicate a 7.41 reduction in RMSE and a 1.2 improvement in VAF values. © 2024 IEEE.

Item Type:	Conference Paper
Publication:	Oceans Conference Record (IEEE)
Publisher:	Institute of Electrical and Electronics Engineers Inc.
Additional Information:	The copyright for this article belongs to the publishers.
Keywords:	Autonomous underwater vehicles; Image thinning; Time difference of arrival, Autonomous underwater vehicles]; Doppler velocity logs; Inertial measurements units; Learning-based approach; Log information; Memory neuron networks; Network-based; Root mean square errors; Underwater environments; Velocity estimation, Neurons
Department/Centre:	Division of Mechanical Sciences > Aerospace Engineering(Formerly Aeronautical Engineering)
Date Deposited:	14 Nov 2024 20:31
Last Modified:	14 Nov 2024 20:31
URI:	http://eprints.iisc.ac.in/id/eprint/86695

Actions (login required)

View Item