In a previous article, we introduced the notion of fleet dispatching algorithms in charge of assigning CCAM vehicles to users when they request a trip. We highlighted that the design of these algorithms represents a major area of research where various techniques are applied but where heterogeneous users, especially vulnerable ones, are almost never considered. Existing algorithms are designed to maximize the overall efficiency of the service with respect to wait times, travel times, and vehicles distances.
In SINFONICA, we investigate inclusivity related implications of CCAM services and how dispatching algorithms can accommodate the presence of vulnerable users. Our first goal is to identify how existing algorithms behave when vulnerable users need more time than non-vulnerable ones for pickup and drop-off. SystemX built for SINFONICA a specific CCAM simulation framework by extending an existing agent-based simulation tool to support vulnerable users’ demand. Each traveller and vehicle are simulated individually and all steps for sending requests, taking operational decisions, and pickup and drop-off interactions are considered.
Using our simulation platform, we choose to investigate two well-documented algorithms that have been studied extensively in the literature for a large variety of use cases. By jointly varying the used dispatch algorithm, the fleet size, the percentage of vulnerable users and the time required by these users for pick-up and drop-off, we performed an extensive analysis and identified the behaviour of the CCAM fleet management services under the presence of vulnerable users. A particular key measure is the rejection rate (the probability for a user’s request being rejected) separately for vulnerable and non-vulnerable users.
The results of our simulations show that existing reference algorithms are not well-suited for handling a mix of vulnerable and non-vulnerable users. Substantially higher rejection rates are observed for vulnerable users than for non-vulnerable ones. Since the algorithms ought to achieve the best wait and travel times for the service users, non-vulnerable ones are systematically favoured when competing with vulnerable users. This effect is magnified when the vehicle interaction time required by vulnerable users increases.
To our knowledge, this study is one of the first to investigate the topic of inclusivity of fleet dispatching algorithms in a systematic and quantitative manner. Our first results shed light on the inclusivity issues raised by existing approaches and identify the need for mitigation measures in order to address these challenges.
Out of this work SystemX will develop effective mitigation solutions within the dispatch algorithms to ensure that the inclusiveness factors will be increased for CCAM operations.
Within SINFONICA, IRT SystemX deploys its expertise on the modelling and simulation of mobility services. It develops a decision support tool to design and assess inclusive CCAM services in territories.
Authors: Tarek Chouaki, Sebastian Hörl, Yann Briand from IRT SystemX.
Keywords: CCAM fleet services, heterogeneous demand, operational strategies, inclusivity.