T2NBS:Planning Night-time Demand-oriented Bus Systems With Urban Computing Approaches

Most governments around the world have noticed the indispensable role of public transport systems in sustainable development of cities,and they have steadily introduced measures to ensure greater convenience and better mobility for the citizens via bus and subway systems.However,for students and workers such as sanitary workers or small business owners,who return home late at night,it is hard to benefit from the pre-existing public transport systems.Most of them suffer from poor working conditions and low salaries,yet they still have to pay the late-night extra charge when taking taxis to return home,and occasionally encounter illegally operated ones.Therefore,as more cities are enjoying the prosperity of the night-time economy,there has reached a consensus amongst citizens on the need for a bus service that operates from late night to dawn.The paper studies the current situation of night-time buses in China,and finds that most of them,essentially a service time extension of day-time buses,turned to be a failure after a limited months or years,as they were unable to avoid the excessive empty-run rate which resulted in operational losses.By contrast,evidence shows that a customized bus system which provides direct and efficient transit services for commuters with similar night-time travel demands might be an appropriate transport format for current Chinese cities to combine citizen’s night-time mobility rights and bus company’s healthy interests.However,new questions have arisen – how can we understand the travel demands of late-night commuters when manual traffic survey is obviously inefficient and costly under this situation,and how can we adjust the proposed bus system to the dynamic travel demands in order to achieve the win-win goals for guaranteed mobility rights and bus profits?Fortunately,urban computing,the emerging methodology for ubiquitous computing of city-wide big data to optimize urban systems,inspires the author with an attempt to utilize the extractive floating car data,i.e.,taxi GPS data here,and OSM transport network data to tackle the above questions.Considering the fact that taxi is almost the only option for intracity travelling when most bus systems are out of service during late night,we could assume that the dynamic night-time travel demands are actually reflected by taxi trajectories.This paper proposes an urban-computing-featured night-time bus planning framework called T2NBS(taxi to night-time bus service),which is applicable to taxi trajectories and other multiple travel data sources.A mathematical programming formulation is put forward to simultaneously optimize bus stop deployment,bus routes,service timetables and passengers’ probability of choosing the nigh-time buses.The paper further develops a heuristic solution framework including a grid-density-based clustering method for discovering potential travel demands efficiently,a bus stop deployment algorithm to minimize the number of stops and the walking distance for commuters to get to the stops,and dynamic-programming-based routing and timetabling algorithms for maximizing estimated profits.The paper conducts an experiment on a small-scale network,i.e.,Sioux Falls network,to verify the performance gap between the optimal solution and the T2 NBS framework.A case study is then conducted on the one-month taxi GPS trajectory data in Xi’an,China.The study demonstrates that the night-time bus lines and timetables generated by T2 NBS could achieve higher profit compared with baseline methods,and they also provide efficient transit services with acceptable walk distance and small departure time adjustments.For the socially and economically disadvantaged,the moderate increase in travel time could be paid off by the significant savings in travel fares.Thus,the proposed win-win situation could be realized via the T2NBS framework.