A systematic review of agent-based models for autonomous vehicles in urban mobility and logistics: Possibilities for integrated simulation models

摘要

This paper provides a systematic review of studies that model future autonomous vehicles (AVs) in urban mobility and logistics using an agent-based model (ABM). Eighty relevant papers published between 2015-2020 are included. This review evaluates the quality of each model through the newly developed AAODD (Autonomous vehicle Agent-based Overview, Design concepts and Details) protocol in four perspectives: model transparency, research questions, network + agent behaviour and model execution. Then, similarities and particularities of the papers are summarised based on whether AVs are used for passenger mobility or for freight logistics.The reviewed mobility papers exhibit the ABMs' applicability and flexibility for studying different AV operational patterns. We summarise the respective modelling specifications in research questions, data collections, model platforms, model executions, and scenario variations. Overall, by combining AVs with different technical possibilities and operational strategies, the simulation results illustrate that AVs would bring vast benefits to urban mobility networks.Unlike AVs' usage in passenger mobility, it is still unclear how AVs will serve future urban logistics demands. Many operational patterns, including unmanned aerial vehicles, autonomous robots, autonomous trucks, and autonomous cargo-hatching vehicles coexist in the research community. The number of relevant papers remains low due to the early stage of AV technology deployment. However, ABMs also prove their applicability for analysing the macro and mesoscopic impacts of freight AV transportation on urban network. The reviewed papers suggest that implementing AVs for urban logistics is more efficient and environmentally friendly than conventional delivery vehicles.Overall, ABMs are a powerful tool for simulating future urban scenarios with AVs. By calculating the average score of reviewed papers obtained via AAODD annually, we find that studies simulating AV systems in ABMs are getting more complex by combing diverse technological possibilities, supporting the simulation of more realistic future scenarios. The outcomes of reviewed ABM research reveals that AVs are a promising solution for solving current urban transport problems. Lastly, as nearly all current reviewed studies focus on AVs' usages in either mobility or logistics, research on an integrated simulation that combines AVs for both purposes has been scarcely addressed. We discuss its necessities, possibilities, and opportunities in this paper, to steer future research in this direction.