Autonomous bus timetable synchronization for maximizing smooth transfers with passenger assignment

Abstract

Autonomous bus (AB), as a major component of public transport systems in the future, are expected to positively impact safety, congestion, private car ownership, energy consumptions, and CO2 emissions. The adoption of ABs will inevitably bring a magnitude of changes and revolutions for the traditional bus operations. This paper investigates the AB timetable synchronization problem (AB-TSP), which involves both the AB timetabling and the passenger assignment. The AB-TSP is first formulated as mixed integer nonlinear programming (MINLP) and then converted to an equivalent mixed integer linear programming (MILP). Two families of valid inequalities are proposed based on the property of the developed model in order to accelerate the solution process. A numerical example on a medium-size AB network shows the effectiveness and efficiency of the valid inequalities, and we also make a detailed comparison between the AB-TSP and the traditional bus timetable synchronization problem (TB-TSP). Finally, a genetic algorithm (GA) framework is designed to cope with the large-scale AB-TSP, and the results of a case study based on the Tower Transit SG bus network show that the intractable real-world problem can also obtain satisfactory solutions within a reasonable computation time.