🤖 AI Summary
This study addresses the performance degradation of underwater monocular visual SLAM (vSLAM) under extreme conditions—including turbid water, low illumination, and varying depth—by systematically characterizing feature matching failure and scale drift induced by underwater scattering media. We propose a physics-informed image enhancement pre-processing paradigm tailored for low-visibility and non-uniform illumination, integrating physically motivated dehazing with adaptive contrast enhancement. Extensive cross-scenario validation is conducted on a real AUV platform and a high-precision laboratory calibration setup, benchmarking against ORB-SLAM2 and other mainstream monocular SLAM methods using ground-truth extrinsic parameters. Results show that monocular SLAM accuracy degrades by 62% on average underwater; our pre-processing improves absolute trajectory error (ATE) by up to 41% and increases map completeness by 3.2×. This work establishes the first quantitative model of underwater vSLAM performance degradation and delivers a transferable, robustness-enhancing framework for practical deployment.
📝 Abstract
Abstract. In this paper, we present a comprehensive investigation of the challenges of Monocular Visual Simultaneous Localization and Mapping (vSLAM) methods for underwater robots. While significant progress has beenmade in state estimation methods that utilize visual data in the past decade, most evaluations have been limited to controlled indoor and urban environments, where impressive performance was demonstrated. However, these techniques have not been extensively tested in extremely challenging conditions, such as underwater scenarios where factors such as water and light conditions, robot path, and depth can greatly impact algorithm performance. Hence, our evaluation is conducted in real-world AUV scenarios as well as laboratory settings which provide precise external reference. A focus is laid on understanding the impact of environmental conditions, such as optical properties of the water and illumination scenarios, on the performance of monocular vSLAM methods. To this end, we first show that all methods perform very well in air and subsequently investigate the degradation of their performance in ever more challenging underwater environments. The final goal of this study is to identify techniques that can improve accuracy and robustness of SLAM methods in such conditions. To achieve this goal, we investigate the potential of image enhancement techniques to improve the quality of input images used by the SLAM methods, specifically in low visibility and extreme lighting scenarios in scattering media. We present a first evaluation on calibration maneuvers and simple image restoration techniques to determine their ability to enable or enhance the performance of monocular SLAM methods in underwater environments.
