Multi-USV cooperative path planning by window update based self-organizing map and spectral clustering

Abstract

Unmanned surface vehicles (USVs) have shown tremendous application prospects recently. In this paper, a hierarchical two-layer framework is proposed for the cooperative path planning of multi-USV system in complex marine environments. The whole framework incorporates the low-level layer planning the optimal path for each USV, and the high-level layer assigning targets for the multi-USV system. First, to avoid collisions and access multiple targets in the optimal sequence, an improved self-organizing map (SOM) based on window update is proposed. The energy cost is calculated to evaluate the path planning results and construct a similarity graph. Then, the spectral clustering method is utilized based on the customized graph to determine the allocated targets which each USV should access in the low-level layer. Finally, a double smoothing strategy based on B-spline is proposed, and the line-of-sight (LOS) guidance method and active disturbance rejection control (ADRC) controller are used to achieve accurate tracking of paths by USV. The simulation results in various scenarios validate the feasibility and effectiveness of our method for multi-USV cooperation, as well as path traceability.