Detecting the borders of a pool-like environment by using scanning sonar data for AUVs

This study presents a novel, cost-effective method for real-time underwater vehicle (AUV) localization in enclosed environments, pools or marinas. Traditional underwater localization techniques, often based on acoustic systems, prove costly and impractical for confined spaces. This study focuses on identifying environmental boundaries and, utilizing this information, solving the AUV\\\'s localization problem by leveraging data acquired from a 360-degree field-of-view sonar scanning sensor mounted on the AUV. Raw sonar data is processed and subsequently clustered using the K-means algorithm, enabling the identification of environmental features such as edges and corners. These identified features are then matched against a pre-existing environment map to determine the AUV\\\'s instantaneous position. Experimental results demonstrate the accuracy and reliability of the proposed approach, with small error values in corner point estimations. While the method\\\'s low computational complexity makes it suitable for real-time applications, the absence of complex and high-cost equipment requirements offers a significant advantage for daily applications. This study suggests that the proposed method has a broad application potential in AUV navigation through its iterative use.