A hierarchical cooperative encirclement strategy for heterogeneous unmanned surface vessels under kinodynamic constraints

This work proposes a hierarchical cooperative encirclement strategy for heterogeneous Unmanned Surface Vessels (USVs) under kinodynamic constraints. The proposed framework integrates dynamic formation generation, global task allocation, and local kinodynamic execution into a unified receding horizon architecture. To improve encirclement efficiency, the task allocation problem is formulated as a Linear Assignment Problem (LAP), where the heterogeneous maneuvering capabilities of different pursuit USVs are explicitly incorporated into the time cost matrix. Furthermore, a Second-Order Dynamic Window Approach (SODWA) is developed for local trajectory planning under underactuated marine dynamics. Unlike conventional DWA methods that sample in the velocity space, the proposed planner directly samples in the actuator thrust space, enabling the generation of smooth and physically executable trajectories while respecting actuator and motion constraints. A discrete time Lyapunov-based analysis is further provided to evaluate the convergence properties of the proposed strategy. Simulation results demonstrate successful encirclement by heterogeneous pursuit USVs across multiple representative target maneuvering profiles. Additional robustness simulations further indicate that the proposed framework satisfies the prescribed encirclement criterion under bounded maritime disturbances and model uncertainties.