DEPART: Dynamic Route Planning in Stochastic Time-Dependent Public Transit Networks

While providing intelligent urban transportation services is one of the key enablers for realizing smart cities, existing transit route planners mainly rely on static schedules and hence fall short in dealing with uncertain and time-dependent traffic situations. In this paper, by leveraging a large set of historical travel smart card data, we propose a method to build a stochastic time-dependent model for public transit networks. In addition, we develop DEPART1-a dynamic route planner that takes the stochastic models of both bus travel time and waiting time into account and optimizes both the speediness and reliability of routes. Experiments on real bus data set for the entire city confirm the quality and accuracy of the routes returned by DEPART in comparison to state-of-the-practice route planners.