Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Transformer-Based Robust Underwater Inertial Navigation in Prolonged Doppler Velocity Log Outages

Zeev Yampolsky, Nadav Cohen, Itzik Klein

TL;DR

The paper tackles prolonged DVL outages in AUV navigation by introducing ST-AidedEKF, which uses ST-BeamsNet to predict missing DVL velocities from past DVL and IMU data and feeds them into an EKF as surrogate DVL updates. The approach preserves continuous velocity updates during outages, achieving up to 63% improvement in velocity RMSE and up to 95% reduction in final position error on real Mediterranean-sea missions. The method is validated on Snapir AUV data, showing strong robustness to outages up to 50 seconds and highlighting the practical impact for reliable autonomous underwater navigation.

Abstract

Autonomous underwater vehicles (AUV) have a wide variety of applications in the marine domain, including exploration, surveying, and mapping. Their navigation systems rely heavily on fusing data from inertial sensors and a Doppler velocity log (DVL), typically via nonlinear filtering. The DVL estimates the AUV's velocity vector by transmitting acoustic beams to the seabed and analyzing the Doppler shift from the reflected signals. However, due to environmental challenges, DVL beams can deflect or fail in real-world settings, causing signal outages. In such cases, the AUV relies solely on inertial data, leading to accumulated navigation errors and mission terminations. To cope with these outages, we adopted ST-BeamsNet, a deep learning approach that uses inertial readings and prior DVL data to estimate AUV velocity during isolated outages. In this work, we extend ST-BeamsNet to address prolonged DVL outages and evaluate its impact within an extended Kalman filter framework. Experiments demonstrate that the proposed framework improves velocity RMSE by up to 63% and reduces final position error by up to 95% compared to pure inertial navigation. This is in scenarios involving up to 50 seconds of complete DVL outage.

Transformer-Based Robust Underwater Inertial Navigation in Prolonged Doppler Velocity Log Outages

TL;DR

The paper tackles prolonged DVL outages in AUV navigation by introducing ST-AidedEKF, which uses ST-BeamsNet to predict missing DVL velocities from past DVL and IMU data and feeds them into an EKF as surrogate DVL updates. The approach preserves continuous velocity updates during outages, achieving up to 63% improvement in velocity RMSE and up to 95% reduction in final position error on real Mediterranean-sea missions. The method is validated on Snapir AUV data, showing strong robustness to outages up to 50 seconds and highlighting the practical impact for reliable autonomous underwater navigation.

Abstract

Autonomous underwater vehicles (AUV) have a wide variety of applications in the marine domain, including exploration, surveying, and mapping. Their navigation systems rely heavily on fusing data from inertial sensors and a Doppler velocity log (DVL), typically via nonlinear filtering. The DVL estimates the AUV's velocity vector by transmitting acoustic beams to the seabed and analyzing the Doppler shift from the reflected signals. However, due to environmental challenges, DVL beams can deflect or fail in real-world settings, causing signal outages. In such cases, the AUV relies solely on inertial data, leading to accumulated navigation errors and mission terminations. To cope with these outages, we adopted ST-BeamsNet, a deep learning approach that uses inertial readings and prior DVL data to estimate AUV velocity during isolated outages. In this work, we extend ST-BeamsNet to address prolonged DVL outages and evaluate its impact within an extended Kalman filter framework. Experiments demonstrate that the proposed framework improves velocity RMSE by up to 63% and reduces final position error by up to 95% compared to pure inertial navigation. This is in scenarios involving up to 50 seconds of complete DVL outage.

Paper Structure

This paper contains 12 sections, 19 equations, 7 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Problem Formulation
  3. DVL Velocity Updates
  4. EKF Formulation for INS/DVL Fusion
  5. Proposed approach
  6. ST-BeamsNet Network Architecture
  7. Set Transformer Aided EKF
  8. Experimental Results
  9. Datasets
  10. Training and Evaluation
  11. Results
  12. Conclusions

Figures (7)

  • Figure 1: A DVL transmits acoustic beams in ""*$\times$ shape configuration.
  • Figure 2: Block diagram of the ST-AidedEKF framework.
  • Figure 3: Block diagram illustrating the ST-BeamsNet architecture which process the DVL and IMU measurements to regress the missing DVL velocity vector due to DVL outage. The input to ST-BeamsNet are the $n$ past DVL measurements, and $m$ past IMU measurements.
  • Figure 4: Snapir AUV being lowered into the Mediterranean sea before an experiment.
  • Figure 5: Improvement $[\%]$ of the average velocity RMSE of ST-AidedEKF over the baseline, for each outage duration. The blue columns represent the results of mission M12, and the orange columns represent the results of mission M13.
  • ...and 2 more figures