Planning for the future of transport with agent-based modelling: The role of on-demand mobility services

Abstract

Today's urgent need for climate change solutions makes it critical to push for more sustainable transport solutions in our cities. Autonomous vehicle technology combined with Mobility on-demand (MoD) services, also known as shared automated vehicles (SAVs), are being considered as a potential way of creating an efficient and sustainable transport system. In particular, ridepooling can combine the benefits of public transport and private vehicles. Agent-based simulation models are extensively used to simulate these systems in order to capture the dynamics between demand and supply. Thus, policies and operational decisions that could directly impact travellers can be modelled and tested to manage travel demand. However, modelling such a system requires rich data and complex strategies that consider the interactions between behavioural parameters and model components that affect the system. This means that there is no perfect simulation approach that can provide a complete model of the real world. Therefore, this dissertation aims to develop and test strategies for effective simulation of on-demand mobility services and to define best practices for using such models, which in turn should enable the creation of useful models that are representative of the real world and valuable for policy planning and forecasting. The studies in this dissertation examine different agent-based simulation strategies for MoD services, addressing issues related to data availability, model richness, decision-making reliability, and the influence of equity policies on MoD simulations. Initially, it extensively discusses the rationale for using agent-based simulations for MoD services. One study focuses on reliable decision-making in agent-based simulation tools such as MATSim, and sheds light on effective ways to provide reliable MoD service-level information to travellers. Another study addresses the need to enrich simulation data with group travel and weekend travel behaviour in MoD travel demand modelling, given their impact on operational decisions such as fleet sizing and vehicle occupancy. A concluding study explores equity considerations in MoD simulations, introducing a framework for incorporating equity as a key evaluation component in MoD service designs. The study provides insights into the nuances of MoD usage across different societal groups and local contexts, highlighting the need for transport planners to adopt an equity-oriented approach when testing for MoD policies.