Urban logistics UAV path planning is a core content of UAV mission planning system. In order to realize the path planning problem of logistics UAVs safely and efficiently, firstly, the environment modeling was carried out by using grid method. Considering the performance limitations of UAVs, the path planning model of multi-constraint logistics UAVs was established by taking the shortest path length, the height variation of UAVs and the minimum grid risk as targets. Secondly, in view of the problems existing in the traditional particle swarm optimization algorithm, Singer mapping was introduced to improve the initial particle distribution, linear adjustment of acceleration coefficients and maximum velocity, new updating strategy of particle position, and dynamic adjustment of inertia weight, and the improved particle swarm optimization algorithm was applied to solve the model. Finally, an example is given for simulation analysis. When the grid size is 5 meters, the path nodes are 5 and the cost function weights are 0.1, 0.4 and 0.5 respectively, compared with the other four algorithms, the total generation value of the proposed algorithm is the best, which is reduced by 44.5%, 3.5%, 42.8% and 30%, respectively. The results show that the model and algorithm in this paper are feasible and effective for UAV path planning.