Coordinating passenger itineraries and vehicle routes in first-mile demand-responsive feeder services

Demand-Responsive Feeder Service (DRFS) is a promising on-demand solution to the first-mile problem, yet its operational performance is often constrained by the conventional process in which itinerary planning and vehicle routing are handled sequentially and independently. This separation can reduce ridesharing opportunities, increase operating costs, and misalign passenger preferences with operator decisions. This study proposes a coordinated operating paradigm where passengers submit multiple candidate boarding and transfer stops and the operator jointly determines itineraries and vehicle routes. To support its implementation and evaluation, we formulate a mixed-integer linear program on a virtual service network and develop a two-stage metaheuristic solver. A case study based on the network in Jiaozhou, Qingdao, shows that the proposed approach reduces passenger cost, operating cost, and weighted total cost by 9.77%, 18.20%, and 13.93%, respectively, compared with a baseline that follows the conventional uncoordinated process. We further find that transfer-stop flexibility contributes more to cost reduction than boarding-stop flexibility, as transfer choices directly affect the feasibility and efficiency of ridesharing.