Underwater autonomous exploration method for flat area based on region growing

This paper presents an autonomous exploration method for underactuated unmanned underwater vehicle (UUV) to search flat areas without the prior maps. To address the poor autonomy of traditional target search methods, a real-time construction of 3D grid maps for navigation is proposed. In terms of searching flat areas, the region segmentation method based on region growing is introduced, and then flat area evaluation methods combining principal component analysis (PCA) and Random Sample Consensus (RANSAC) algorithms are employed to select flat areas. During the exploration process, the boundary-based exploration method is used to generate target points and exploration points are selected according to the suitable evaluation function, along with the Rapidly-Exploring Random Tree Star (RRT*) algorithm and the dynamic window approaches (DWA) for exploration path planning. Finally, the effectiveness of the underwater autonomous exploration method in searching flat areas is validated through practical simulation studies.