Research on Cooperative Search and Dynamic Target Encirclement of Unmanned Surface Vehicle Swarms Based on Improved Pheromone Algorithm

Concentrating on the practical needs of effective collaborative search and encirclement techniques for unmanned surface vehicles (USVs). This study proposes a collaborative search technique, focusing on enhancing the efficiency, preventing early pheromone accumulation, and achieving local optimality, in response to challenges in the regional coverage search of USVs in unfamiliar settings. The system creates a technique for revising hazardous grids and identifying adjacent locations. Global guidance is provided by the primary grid, while the sub-grid serves the purpose of local optimization. The efficiency of searching sub-grids is enhanced by dynamically modifying their search density via the mechanism of pheromone decay or enhancement. A method of dynamically distributed encirclement is suggested to address the challenge of dynamically allocating encirclement roles and preserving the stability of formation during the encirclement of dynamic targets. Encirclement occurs through the creation of encirclement contracts and the creation of potential points for encirclement, while the potential points for encirclement in dynamic targets are judiciously distributed, facilitating rapid formation of an encirclement by several USVs. Following this, simulations are conducted to test the aforementioned algorithms, leading to the optimization of their control parameters. A total of six USVs were employed to conduct a search for coverage with in a 400 m $\times 400$ m zone. The outcomes of the simulations revealed that a single USV located the target in 53.5 seconds, while the rest ceased their search swiftly, created an encirclement at 85.3 seconds, attained the threshold in 101.3 seconds, autonomously avoid obstacles, and finished the encirclement task. Ultimately, the lake study was confirmed using four USVs, which located the target upon achieving 64% area coverage, smoothly transitioned from search to encirclement mode, and ultimately accomplished the encirclement task. The effectiveness and rationality of this method were successfully confirmed through simulations and experiments.