Large-scale activity-based demand generation modeling: A literature review and exploration of potential approaches

Large-scale, activity-based microscopic transport models provide a powerful framework for analyzing dynamic travel demand and assessing the impact of transportation policies on daily travel behavior. At the core of these models is the generation of travel demand, which necessitates the creation of detailed synthetic populations and daily travel plans associated with personalized activities. These time-dependent travel plans are then integrated into dynamic traffic assignment models to simulate agent-based systems. This paper presents a comprehensive review of existing demand generation models within activity-based frameworks, focusing on various methodologies including constraint-based, utility-based, rule-based, learning-based, and hybrid approaches. A comparative analysis is offered, highlighting their theoretical foundations, data requirements, key outputs, and particularly their applications in large-scale microsimulations. In addition, the paper discusses the possibility of collecting input data for these models, as well as explores innovative approaches capable of modeling daily mobility patterns as sequences of activities linked by trips, offering greater flexibility in capturing dynamic travel behavior. Furthermore,potential research directions are also discussed, including the development of travel models for large-scale scenarios using big data sources and the optimization of their integration with dynamic traffic assignment. These methods hold significant promise for integration into large-scale, microscopic dynamic traffic assignment platforms. This study provides critical insights for researchers and practitioners focused on advancing large-scale microscopic traffic modeling to improve decision-making processes.