Abstract: Aiming at the problems that  traditional Artificial Fish Swarm Algorithm (AFSA) is easily falling into the local optimal solution and the planned optimal path is too long in the process of pasture inspection path planning, an improved Artificial Fish Swarm Algorithm integrated with Genetic Algorithm (GA) is proposed to obtain a Genetic Fish Swarm Algorithm (GFSA). On the basis of AFSA, GFSA accelerates the convergence speed and accuracy during the solving process by optimizing the vision and step length. Additionally, the algorithm avoids premature convergence around local optimal solution during iterations by designing a segmented crowding factor and incorporating mutation operations from Genetic Algorithm (GA). Genetic Fish Swarm Algorithm (GFSA) was implemented on a pasture inspection robot for experimental validation. The results showed that in ablation experiments, the path lengths planned by GFSA were consistently shorter than those of the comparison algorithms, with a median optimal path length of 23.2 m. In multi‑algorithm comparison experiments, GFSA achieved the shortest paths and smaller turning angles compared to the other algorithms. In path planning on maps with different obstacle rates, GFSA's initial value and growth slope of the optimal path length were both lower than those of the comparison algorithms, demonstrating better robustness and adaptability.

Key words: ranch, inspection robots, Genetic Algorithm (GA), Artificial Fish Swarm Algorithm(AFSA), path planning

摘要： 针对传统人工鱼群算法（AFSA）在牧场巡检路径规划的过程中易陷入局部最优解、规划的最优路径较长等问题，提出一种改进人工鱼群算法融合遗传算法(GA)得到的遗传鱼群算法（GFSA）。GFSA在AFSA基础上通过优化视野和步长，加快算法在求解过程中的收敛速度和精度；通过设计分段式拥挤度因子、引入遗传算法中的变异操作，避免算法在迭代过程中过早聚集在局部最优解周围。将GFSA搭载在牧场巡检机器人上进行试验验证。结果表明，在消融试验中，GFSA规划的路径长度均低于对比算法，其最优路径长度中位数为23.2 m；在多算法对比试验中，与对比算法相比，GFSA的路径最短且转弯幅度小；在不同障碍物率地图的路径规划中，GFSA的最优路径长度初始值、增长斜率均小于对比算法，具有更好的鲁棒性和适应性。

关键词: 牧场, 巡检机器人, 遗传算法, 人工鱼群算法, 路径规划