Abstract: Autonomous underwater robots (AUVs) can perform tasks such as resource exploration and target search in deep sea areas. At offshore sites, we often face the need to quickly respond to mission deployment and conduct path planning for AUV search and mapping missions. This paper proposes an AUV multi-target search method that clusters first and then searches in the AUV multi-suspect target point search task. It performs K-means clustering on multiple target points in the large-scale sea area to be measured to form multiple target points to be tested. The measurement area reduces the search for non-target areas and ensures higher search efficiency when the AUV carries limited energy; the comb path scheme is implemented in multiple areas to be measured to form an AUV search path in multiple areas. An AUV multi-target clustering search software based on Tkinter was developed to implement the method proposed in this article through visual interactive tools, solving the problems of time-consuming and error-prone manual planning and improving the efficiency of AUV deployment. Finally, We compare the length of AUV paths between different cluster numbers to determine the feasibility and application value of this method.

Key words: autonomous underwater vehicle, multitarget point search, K-means clustering algorithm, Tkinter