UAV path planning has important application value in tasks such as post disaster emergency communication. This article proposes an improved differential evolution - spherical particle swarm optimization algorithm (DE-SPSO), which integrates three-dimensional spatial attitude constraints into particle updates through vector spherical coordinate system encoding, enabling the algorithm to obtain optimal solutions in early iterations and improving convergence speed; At the same time, the dynamic differential evolution mechanism is introduced, which automatically triggers the generation of new particles when the particle is detected to be stuck, strengthening the algorithm’s capability to overcome local optimal solutions. In addition, our algorithm also combines local exploration strategies and path compression techniques to expand the search range, and finally smooths the path through cubic B-spline curves. Simulation experiments show that compared with PSO, SPSO, and SDPSO algorithms, DE-SPSO can obtain better paths faster and exhibit better robustness.