Public transport routing including fixed schedule, shared on-demand and door-to-door services

Abstract

Modelling realistic public transport route choices is a key element for transport planning tools. Although these decisions have been modeled based on travel time and travel cost, the emergence of new public means of transport makes the problem more challenging. In this work, we extend the public transport router of a widely used mobility simulation platform called MATSim. Besides fixed scheduled services, the proposed router is able to include on-demand shared services as well as private door-to-door services. With the co-evolutionary algorithm of MATSim, this router records waiting and travel times from previous iterations to improve public transport routes. This allows to take into account interactions between agents when routes are decided. The approach is tested in a model of a future neighborhood of Singapore, where on-demand services complement the existing fixed schedule supply.