An online path planning algorithm for autonomous marine geomorphological surveys based on AUV

This paper proposed a data-driven bi-pattern (DDBP) path planning algorithm for ocean geomorphological surveys based on Autonomous Underwater Vehicles (AUVs). When an AUV conducts surveys in unknown areas, it uses the observation data of real-time side-scan sonar to conduct environment modeling to drive independent online path re-planning (PRP) according to the feature density of the interesting targets. Based on the DDBP algorithm, the AUV can autonomously focus on regions with rich target distribution and deviate from regions with sparse target distribution without prior knowledge of the task region. The quality and efficiency of the AUV-based surveys can be improved by focusing on the underwater detection area with high feature density. The DDBP algorithm includes two patterns: rough and fine scan, and the corresponding planning pattern is selected according to the distribution of the detected targets. AUV performs online PRP in the corresponding pattern according to the pre-identified strategy set. We conducted simulation experiments and selected sand waves and fish reefs as natural and artificial structures to conduct typical marine survey tests. Compared with the traditional marine survey method, the survey efficiency was increased by 33.6% and 29.6%, respectively, in the two DDBP survey experiments for sand waves; the efficiency increased by 32.9% and 36.7%, respectively, in the two groups of DDBP survey experiments on artificial reefs. The proposed general technical framework for online path planning driven by real-time observation data has good application prospects in underwater archaeology, rapid understanding of specific targets on the seafloor, and search of specific targets.