For three-dimensional UAV route planning in complex environment, the probabilistic particle swarm optimization (PSO) has difficulty in acquiring optimal path or even feasible path due to high dimensionality, while the deterministic sparse A algorithm is likely to be trapped into local search which leads to long search time and expensive computation. Based on hierarchical path planning, the high dimensional path planning could be decomposed to some low dimensional problems. Firstly, some navigation waypoints was planned using PSO, and then the path linking the navigation points was planned using sparse A algorithm. The simulation results indicate the proposed method could improve feasible path planning efficiency in complex environment.