On Accessibility Fairness in Intermodal Autonomous Mobility-on-Demand Systems

Most applied research exploring justice in the domain of transport has focused on the equity evaluation of existing systems. At the same time, most research on transport planning has been implementing conventional utilitarian paradigms, e.g., minimizing the average travel time of the population, without accounting for fairness. This paper aims to bridge this gap and adds to this literature in two ways: by exploring to what extent the application of different justice principles can enhance the fairness of the transport system; and by focusing on realizing such principles in the operation of transport systems rather than merely assessing a given system design. We use an intermodal Autonomous Mobility-on-Demand (AMoD) system as our case study, where a fleet of centrally controlled self-driving cars provides on-demand mobility synergistically with public transit and active modes (biking and walking). We explore how its operation can improve the situation of users that do not own a car. We first formally define a set of justice metrics that differ in terms of distributive principle and the good of concern. The metrics include: minimization of average travel time for the car-less population (i.e., a population-specific application of utilitarianism); avoidance of unacceptably long travel times for the car-less population in line with a sufficientarian approach; and delivery of reasonable travel times to a sufficient set of destinations. We showcase our framework in a real-world case-study in the city of Eindhoven, the Netherlands. Our results show that, compared to conventional utilitarian minimum-travel-time planning, it is possible to significantly improve the situation of the car-less users without affecting conventional performance metrics such as average travel time. Whilst the differences between the proposed sufficientarian deployment models are rather modest, they highlight intrinsic crucial trade-offs that require further consideration and analysis. Overall, these results underscore the importance of taking a transdisciplinary approach addressing planning problems from conceptualization to modeling and optimization in transport and mobility.