In order to improve the operational efficiency of responsive feeder transit, an idea of method for responsive feeder transit system was proposed. Which consists of the coordinated operation of multiple depots and transfer stations, the coordinated design of running routes, and the vehicle scheduling. Considering the mixed demand of reservation and real-time, the two-stage model of responsive feeder transit system was built to coordinate the running routes and vehicle scheduling. The multi- chromosomes genetic algorithm was then designed. The results show that, under the same conditions of mixed demand, the total cost, the running time and the number of departures of the collaborative operation mode is reduced by 14.4%, 24.2%, and 3 times respectively, compared with the independent operation. By increasing the proportion of small vehicle to 50%, the total system cost of the independent operation and the collaborative operation is reduced by 7.3% and 7.5% respectively.