In order to overcome the demerits of traditional city air terminals (CAT), a brand-new service mode based on urban mobile stations (UMS) is introduced in this article for providing the luggage check-in service in urban area. Different from CAT, for each service time period, the UMSs are located based on the real-time passenger distributions. The objectives are to improve the service quality and enhance the competitiveness of air transportation. First, the system components and operation mode of UMS are described. Second, a mathematical model is set up for locating UMSs and assigning passengers to service stations. Three aspects, including the average path length from passengers to UMSs, the maximum acceptable path length for passengers, and the maximum service capacity of a station are considered. Third, to meet the strict time requirement of the UMS system, an effective evolutionary algorithm for optimizing UMS locations in real time is further developed. In the method, the ripple-spreading algorithm is applied for solving the many-to-many path optimization problem and a two-stage adaptive genetic algorithm is utilised for efficiently locating UMSs. Finally, a test case is set up based on the city centre of Tianjin, China, and passenger source locations are obtained by surveying passengers at Tianjin Airport. The reported method is then applied to dynamically locate UMSs in Tianjin urban area and it is compared to some other existing methods to show its effectiveness and efficiency. The service quality of UMS is compared to CAT in both cases with realistic data and randomly generated data to show its advantages.