A Path Planning Approach for Multi-AUV Systems with Concurrent Stationary Node Access and Adaptive Sampling

Stationary node data retrieval and adaptive sampling are two important tasks in marine environment monitoring based on autonomous underwater vehicles (AUVs). Enabling AUVs to carry out these two heterogeneous tasks within a single deployment is crucial for the efficient utilization of hardware resources. This letter presents a path planning method for a multi-AUV system that concurrently performs stationary node access and adaptive sampling. The proposed method assigns nodes for AUVs by dividing the optimal giant route, then partitions the mission area into sampling regions using Voronoi diagrams and assigns tasks to each AUV. Modeling the interesting field as a Gaussian process, each AUV plans its sampling path within the assigned region based on spatial sampling, and the planning objective is to maximize the total covariance of the sampling positions. The proposed planner is compared to existing methods through simulations, and it outperforms the single task-focusing methods by 45.3% and 54.2% in different aspects, effectively enabling the AUVs in accomplishing the two heterogeneous tasks concurrently. The real-world test has further substantiated the planner's feasibility.