Boundary adaptive and dynamic neural network dual layer path planning algorithm for polar multi-mission area under-ice feature scanning

Autonomous Underwater Vehicles play a crucial role in polar under-ice feature scanning, contributing to explore ecological changes in polar regions. However, multi-mission under-ice feature scanning face significant challenges due to complex boundaries, energy constraints, and environmental uncertainties. To address these issues, this paper proposes a Boundary Adaptive and Dynamic Neural Network Dual Layer path planning algorithm, which consists of coverage path planning and connectivity path planning. First, Boundary Adaptive Coverage based on Dynamic Grid algorithm is introduced, which optimizes coverage paths by adaptively inserting boundary waypoints to improve coverage efficiency of AUV. Second, Dynamic Neural Network based on Ocean Current Energy Consumption algorithm is proposed, which integrates ocean current effects to save energy and real time replan connection paths. Simulation results demonstrate that it outperforms current methods in terms of total path length, turning times, and energy consumption, and its practicality is verified by “Xing Hai 1000” AUV field experiments.