Multimodal passive tracking of passengers to analyse public transport use

Abstract

We tackle the problem of understanding multimodal passenger flows to better address mobility. In particular, we use large scale, long-term GPS passive tracking of travellers, to improve public transport operations in urban areas. We combine this with realised operation data from a transit operator. We develop an unsupervised tool able to detect passengers’ travel behaviour and the exact public transport means they took, without any assumption on the frequency of GPS data. Furthermore, we show the significant advantage of using past user data to improve the detection algorithm, in particular to identify the points where users transfer. We test this approach in Zurich and refer to the multimodal mobility supply available there. Ultimate goal of this ongoing work is to understand from realised mobility how passengers react to disturbances, especially for working days and peak hours, and to the mitigation action undertaken by operators in case of disturbances, in order to design good and effective mitigation actions.