Multimodal and Multicriteria Path Set Computation for Journey Planning in Time-Dependent Mobility as a Service Networks

Abstract

Technological advancements and the rise of the sharing economy have led to the emergence of the Mobility as a Service (MaaS) concept. In MaaS, traditional and new mobility services are integrated and offered to users as a seamless travel experience. The wide variety of mobility services in MaaS, their inherently dynamic service attributes and the multifactorial MaaS users's journey choice process render optimal journey planning a challenging problem. This paper formulates and solves the fully dynamic, multimodal and multi-criteria path set computation problem for MaaS journey recommendation applications. Specifically, a generalized time-dependent multimodal and multi-attribute supernetwork model is adopted, enabling realistic integration of traditional transport and new mobility services, representation of their functional characteristics and modelling of several journey attributes. Two new generic algorithms, based on backward label-setting and label-correcting approaches, are designed to generate exact and heuristic Pareto MaaS journey sets. Algorithm applications are compared with each other as well as with traditional forward-propagating approaches to prove their optimality, result quality and efficiency. Our methods are tested for a prototypical virtual city and insights are extracted with regards to computational performance, path quality and multimodality aspects. Results indicate that the heuristically enhanced label-correcting approaches outperform corresponding label-setting ones in terms of computational performance, while keeping the same quality levels. Exactforward-propagating approaches, albeit non-optimal by design, may also yield acceptable quality solutions more efficiently. Also while both approaches are capable of producing diverse and high quality Pareto sets when heuristically enhanced, the impact on multimodal path computation may be significant.

Keywords: Mobility as a Service, Multimodal, Multi-criteria, Time-dependent, Journey Planning, Supernetworks