To address the issues of low optimization accuracy, slow convergence speed, and susceptibility to local optima encountered by traditional Artificial Fish Swarm Algorithm (AFSA) when solving the Traveling Salesman Problem (TSP), this paper proposes an improved AFSA algorithm integrated with cross-over mutation. Firstly, by introducing cross-over mutation operations during the iterative solving process of the fish swarm, population diversity is enhanced, thereby improving the algorithm's capability to find better solutions in global search. Secondly, an adaptive fish swarm strategy is introduced, dynamically adjusting the visual range and crowding factor to enhance the algorithm's local exploration capability and convergence speed. Thirdly, simulation verification is conducted using the TSPLIB dataset in the MATLAB environment. Results demonstrate that the improved AFSA algorithm exhibits significant improvements in convergence speed and optimization accuracy compared to traditional methods, with enhanced ability to escape local optima and path planning results closer to the optimal solution. Finally, further improvements are made to the classical TSP model in terms of map dimensions and paths, ultimately realizing the application of this improved algorithm in three-dimensional multi-point coverage path planning.