Towards Versatile Opti-Acoustic Sensor Fusion and Volumetric Mapping for Safe Underwater Navigation
Ivana Collado-Gonzalez, John McConnell, Brendan Englot
AI summary
Problem
Vision-based perception fails in turbid waters while sonar suffers from low resolution and elevation ambiguity, leaving a gap for robust 3D mapping in complex, variable-visibility underwater environments.
Approach
The framework projects stereo sonar features onto camera-segmented regions of interest to resolve elevation ambiguity, assigns reliability scores to each measurement, and fuses them using a confidence-weighted Gaussian Process volumetric mapper.
Key results
- First stereo sonar and monocular camera fusion strategy for underwater sensing
- Confidence-weighted Gaussian Process mapping prioritizes reliable measurements
- Tank experiments validate accurate reconstruction of complex geometries
- Field tests demonstrate robust adaptability to low-visibility marina conditions
Why it matters
Provides AUV operators and researchers with a reliable, open-source mapping tool that maintains navigation safety regardless of water clarity or lighting conditions.
Abstract
Accurate 3D volumetric mapping is critical for autonomous underwater vehicles operating in obstacle-rich environments. Vision-based perception provides high-resolution data but fails in turbid conditions, while sonar is robust to lighting and turbidity but suffers from low resolution and elevation ambiguity. This paper presents a volumetric mapping framework that fuses a stereo sonar pair with a monocular camera to enable safe navigation under varying visibility conditions. Overlapping sonar fields of view resolve elevation ambiguity, producing fully defined 3D point clouds at each time step. The framework identifies regions of interest in camera images, associates them with corresponding sonar returns, and combines sonar range with camera-derived elevation cues to generate additional 3D points. Each 3D point is assigned a confidence value reflecting its reliability. These confidence- weighted points are fused using a Gaussian Process Volumetric Mapping framework that prioritizes the most reliable mea- surements. Experimental comparisons with other opti-acoustic and sonar-based approaches, along with field tests in a marina environment, demonstrate the method’s effectiveness in captur- ing complex geometries and preserving critical information for robot navigation in both clear and turbid conditions. Our code is open-source to support community adoption.