Towards Effective Visual-Inertial SLAM with Passive-Only Sensors for Low-Cost Autonomous Underwater Vehicles

Grant Schwidder; David Widhalm; Junaed Sattar

cs.ROSep 17, 2026

Towards Effective Visual-Inertial SLAM with Passive-Only Sensors for Low-Cost Autonomous Underwater Vehicles

Authors: Grant Schwidder, David Widhalm, Junaed Sattar

Organizations: Department of Computer Science & Engineering, University of Minnesota–Twin Cities, Minneapolis, MN 55455, USA

Abstract

Improvements to Visual-Inertial Simultaneous Localization and Mapping (VI-SLAM) for low-cost autonomous underwater vehicles (AUVs) are critical for transitioning advanced marine robotics from specialized labs to broader research and hobbyist applications. While high-end AUVs typically rely on expensive sensor suites - such as Doppler Velocity Logs (DVLs) and Ultra-Short Baseline (USBL) systems - this work demonstrates that robust, high-quality navigation is achievable using a sub-$10, 000(USD) platform equipped only with inexpensive consumer-grade sensors. By leveraging a similarly priced, open-source AUV, we evaluate the performance of stereo cameras, Micro-electromechanical System (MEMS)-based IMUs, and depth sensors in a fully unconstrained 6-degree-of-freedom (6-DOF) underwater environment. We analyze the efficacy of off-the-shelf SLAM packages and propose optimizations for sensor fusion to mitigate the visual and physical challenges of untethered underwater operation. Our results prove that a usable SLAM solution can be accessible to the masses, providing a benchmark for expectations in demanding, real-time maritime missions without the financial barrier of industrial-grade hardware.

Figures & tables

Panel of Figure 1 — Fig. 1 : A snapshot of VI-SLAM running on board our low-cost AUV on an ocean reef. Fig. 1(a) shows actual images acquired by the on-board stereo camera and visual features used for SLAM. Fig. 1(b) shows a point cloud representation of the reef reconstructed aftwerward.

Panel of Figure 2 — Fig. 2 : Evolution of tracking performance: (a) Absolute VIO showing significant shearing; (b) Differential integration improving consistency; and (c) Full EKF fusion incorporating IMU and pressure data to eliminate drift and handle visual degradation. All implementations are running with RTAB-Map loop closures enabled.

Sensor	Pose			Orientation			Linear Vel. ( $v$ )			Angular Vel. ( $\omega$ )			Acceleration ( $a$ )
	$x$	$y$	$z$	$\phi$	$\theta$	$\psi$	$\dot{x}$	$\dot{y}$	$\dot{z}$	$\dot{\phi}$	$\dot{\theta}$	$\dot{\psi}$	$\ddot{x}$	$\ddot{y}$	$\ddot{z}$
ZED VIO	✓	✓	✓
ZED IMU				✓	✓	✓				✓	✓	✓
2nd IMU				✓	✓	✓				✓	✓	✓	✓	✓	✓
Pressure Sensor			✓

TABLE I : EKF configuration: active state variables per sensor

Fig. 3 : Layout of the coordinate frame configuration of the AUV, omitting the world frame. The $r_{p}$ vector shows the distance used in the Transport Theorem.

Panel of Figure 5 — Fig. 4 : Comparison of map generation in an indoor office office environment with differences in turn velocity. Higher turn velocity yielded more drift in the 90 degree angles of the hallway, though the map remains relatively accurate to the ground truth. The slow turns resulted in a high quality reconstruction. These results demonstrate that the external IMU greatly improves the orientation in the state estimate.

Fig. 5 : Trajectory of AUV in an indoor office office environment overlaid with a ground truth floor-plan. The trajectory is accurate without any constraints imposed on the assumed movement of the AUV. Notably, the depth sensor keeps the state estimate grounded, preventing vertical drift and ensuring the path is aligned to the 2D floor plan.

Panel of Figure 7 — Fig. 6 : Mapping comparison of pool environment with and without loop closure. With loop closure disabled, a significant gap exists between the start and end position. Loop closure resolves the gap but sacrifices the accuracy of the pool’s rectangular geometry. This highlights the inherent challenge of using vision for odometry in environments with repetitive textures.

Configuration / Scenario	RMSE Value
Hallway (Slow Turns)	$0.56$ m
Hallway (Fast Turns)	$0.71$ m
Pool With Loop Closure	$1.04$ m
Pool With No Loop Closure	$1.14$ m

TABLE II : RMSE Results of our VIO-SLAM approach from different environments

Fig. 7 : Generated point cloud map of a coral reef. The map shows an accurate path with a distinguishable floor and coral wall. This highlights the ability to obtain quality maps from unconstrained environments, suitable for later analysis of seafloor ecosystems.

Panel of Figure 10 — Fig. 8 : Continuous pose estimation during ocean navigation (top), containing images from the traversed path (bottom). The overlaid path illustrates a smooth state estimate despite environmental constraints. This lack of discontinuities underscores the robustness of the integrated sensor configuration in handling non-linear dynamic data.

Visual-Inertial SLAM

Visual-inertial simultaneous localization and mapping (SLAM) combines camera and inertial data to estimate motion and build consistent maps.

See all 29 papers

APVI-SLAM: Real-Time Acoustic-Pressure-Visual-Inertial Localization and Photorealistic Mapping System in Complex Underwater Environment
Jul 7, 2026Hanwen Zhang, Yipeng Zhu, Xiaopeng Guo +2Underwater Robotics Visual-Inertial SLAM
Division of Integrative Systems and Design, Hong Kong University of Science and Technology · School of Interdisciplinary Science, Beijing Institute of Technology · Division of Integrative Systems and Design, the Department of Computer Science and Engineering, and the Department of Ocean Science, Hong Kong University of Science and Technology
DIVO: Continuous-time DVL-Inertial-Visual Odometry for Unmanned Underwater Vehicles
Jul 6, 2026Kyungmin Jung, Angad Bajwa, Junha Yoo +2Underwater Robotics Visual-Inertial Odometry
Department of Mechanical Engineering, McGill University, Montreal, QC, Canada
SVIn+: Multi-Modal Framework for Robust Underwater State Estimation using 3D Sonar, Visual-Inertial, Water-Pressure, and DVL
Oct 27, 2026Chinmay Burgul, Xiaoming Zhao, Yewei Huang +4Robot State Estimation Underwater Robotics
Mechanical Engineering Department, University of Delaware, Newark, DE, USA. · Dartmouth College, Hanover, NH, USA · Aquaculture Robotics and Automation Group, SINTEF Ocean

Jul 7, 2026J/K move · Enter open · S save

APVI-SLAM: Real-Time Acoustic-Pressure-Visual-Inertial Localization and Photorealistic Mapping System in Complex Underwater Environment

Hanwen Zhang, Yipeng Zhu, Xiaopeng Guo, Huajian Huang+1

Division of Integrative Systems and Design, Hong Kong University of Science and Technology · School of Interdisciplinary Science, Beijing Institute of Technology · Division of Integrative Systems and Design, the Department of Computer Science and Engineering, and the Department of Ocean Science, Hong Kong University of Science and Technology

Underwater Robotics Visual-Inertial SLAM Multimodal Sensor Fusion

Open paper AskPDF arXiv

Extreme subsea environments often cause severe feature de-gradation and estimator divergence in underwater visual-inertial SLAM. Although sensors like Doppler Velocity Logs (DVL) and pressure gauges provide auxiliary constraints, robust multi-sensor fusion during intermittent visual failure remains challenging. To address this, we present APVI-SLAM, a real-time multi-sensor fusion SLAM system that achieves both accurate underwater localization and photorealistic mapping. Our approach introduces a reliability-aware localization framework that dynamically reweights sensor estimators and employs a sliding-window freezing strategy to recover from tracking failures, substantially enhancing system robustness. Furthermore, for high-fidelity scenes reconstruction, we propose an efficient quadtree-guided mapping module that facilitates incremental water-medium modeling and 3D Gaussian optimization. Recognizing the lack of benchmark for underwater mapping evaluation, we also contribute a coral reef surveying dataset with synchronized multi-modality data. Extensive experiments on public and our proposed benchmarks demonstrate that APVI-SLAM achieves state-of-the-art localization and reconstruction quality at real-time speeds.

Abstract

Visual-Inertial SLAM

Explore similar work

APVI-SLAM: Real-Time Acoustic-Pressure-Visual-Inertial Localization and Photorealistic Mapping System in Complex Underwater Environment

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

SVIn+: Multi-Modal Framework for Robust Underwater State Estimation using 3D Sonar, Visual-Inertial, Water-Pressure, and DVL