This paper considers a network scenario assisted by unmanned aerial vehicles (UAVs). In a complex environment with dense obstacles, the UAVs are deployed to provide downlink services to users. Users are located in areas with dense obstacles. Based on the transmission model, the target position of path is determined, this method is used to seek optimal path under constraint conditions. This work proposes the ASPSO (Adaptive spherical vector-based Particle Swarm Optimization) algorithm. Firstly, an initialization way is designed. The method sets different parameters for scene requirements, effectively shortens the initial times. Secondly, differential evolution is introduced during the search process, and a multi-strategy optimization method is proposed. Increasing the search space in the early stages of iteration is beneficial for get rid of the local good solution. In the late stages of iteration, small disturbance is introduced to continue exploring in the neighborhood space of high-quality solution. Finally, a way for path improvement with virtual control points is proposed to smooth the trajectory and reduce fitness. In this paper, the method is compared with PSO (Particle Swarm Optimization) and SPSO (Spherical Vector-based PSO), the ASPSO can quickly obtain high quality initial solutions and has better exploration ability in complex three dimensional environments.