DIVO: Continuous-time DVL-Inertial-Visual Odometry for Unmanned Underwater Vehicles

2026-07-05

AI Summary

This work addresses the degraded accuracy and poor robustness of visual localization and mapping for underwater vehicles under severe light attenuation and suspended particle interference. To this end, the authors propose a continuous-time Gaussian process-based framework for fusing asynchronous, multi-source sensor data. They introduce, for the first time, continuous-time Gaussian processes into underwater odometry by integrating Doppler Velocity Log (DVL), stereo cameras, and an IMU. The approach incorporates a short-term visual association mechanism robust to underwater disturbances and a learned visual front-end that enables adaptive feature extraction and matching without requiring environmental reconfiguration. Evaluated on real-world underwater inspection datasets, the method outperforms existing visual-inertial and acoustic-visual-inertial SLAM approaches in terms of localization accuracy, robustness, and trajectory coverage.

Abstract

This paper presents a novel acoustic-visual-inertial odometry solution leveraging a continuous-time trajectory estimation framework for unmanned underwater vehicles. Underwater environments present unique challenges for visual localization and mapping, such as light attenuation, illumination variance, and the presence of particulate matter. This motivates the use of additional sensing modalities and a visual tracking pipeline that is robust to diverse subsea conditions. The proposed system is the first continuous-time trajectory estimation framework based on Gaussian processes to fuse asynchronous measurements from a Doppler velocity log, a stereo camera, and an inertial measurement unit. Additionally, a novel visual frontend is proposed, incorporating learning-based feature extraction and matching that is robust to the specific challenges that subsea environments present. The proposed framework enables seamless integration of additional sensor modalities in continuous-time and is adaptable to different environments without reconfiguration. The proposed system is extensively tested on real-world underwater inspection datasets, where it outperforms state-of-the-art visual-inertial and acoustic-visual-inertial SLAM algorithms in accuracy, robustness, and trajectory coverage. Notably, the proposed system outperforms the state-of-the-art despite only forming short-term visual data associations.

Problem

• underwater localization
• visual-inertial odometry
• Doppler velocity log
• continuous-time trajectory estimation
• subsea environment

Innovation

• continuous-time trajectory estimation
• Gaussian processes
• acoustic-visual-inertial odometry
• learning-based visual frontend
• asynchronous sensor fusion