Under the background of the rapid development of global intelligent manufacturing, improving the efficiency of intelligent harvesting robots in agricultural operations has become a key research focus. Addressing the complex multi-target traversal problem faced by orchard harvesting robots, this study proposes a multi-target global path planner connecting Particle Swarm Optimization (PSO) and the A* algorithm. First, the A* algorithm is used to calculate the actual collision-free path distance between target points to construct a real cost matrix, overcoming the limitation in traditional Traveling Salesperson Problem (TSP) solutions that rely solely on Euclidean distance and ignore the impact of obstacles. Subsequently, an improved inertia-free discrete PSO algorithm is utilized to solve the multi-target path planning problem and generate the optimal target visit sequence. Finally, the A* algorithm is used to connect the sequence into a complete closed-loop working trajectory. Gazebo simulation tests demonstrate that the proposed algorithm converges efficiently in multi-target scenarios of various scales, significantly improving the global planning quality and computational efficiency of robots in complex orchard environments.