Because of the passenger travel time, vehicle capacity and other restrictions, the larger service area may need to open multiple lines and run multiple vehicles, so it is necessary to further classify the service area, and determine the time of departure, allocate the vehicle, optimize the path by partition. The travel demand of the reserved passenger is considered, and an iterative algorithm is constructed to optimize the path and scheduling in every partition. For each partition, with the passenger travel time window, vehicle capacity, vehicle travel time as constraints, a route optimal model is set up to minimize total vehicle costs based on VRP method, at the same time, a genetic simulated annealing algorithm is designed. Through multiple experiments, the results can be seen, such as: 1) when the vehicle is optional, the total cost can be reduced regardless of the demand project size, but the larger the demand, the lower the total cost; 2) the total cost is related to the number of partitions while low demand can have adverse effects. In higher demand, the appropriate partition will greatly reduce the total cost, and optimal partition number is not increased with increasing demand size; 3) the total cost is significantly reduced when the coordination optimization is used.