Reconfigurable Line Light Sources’ Sensing Method for Vision-Based AUV Guidance

Autonomous underwater vehicles (AUVs) need precise underwater positioning for tasks such as formation, docking, and recovery. In practical applications, visual guidance systems should integrate multiple types of markers (such as points, lines, and QR codes) to compensate for the limitations of single-marker solutions. A reconfigurable marker based on line light sources is first designed, whose morphological transformation is achieved by adjusting its emission angle according to the relative distance. To address the problems of low marker brightness underwater and inconsistent enhancement algorithms, adaptive image enhancement is realized through the combination of the recursive least-squares (RLS) algorithm and multiscale retinex with color restoration (MSRCR) technology. During the extraction of key points from the enhanced images, hierarchical segmentation-based line fitting is implemented by utilizing the optical characteristics of line light sources. Visual positioning under partial feature loss is accomplished by acquiring key points through fit lines. Experimental studies on reconfigurable visual markers are conducted to verify the feasibility and accuracy of the proposed algorithm. Comparative experiments with existing markers are performed to evaluate their respective perception ranges.