Abstract

In transportation system demand modeling and simulation, agent-based modelsand microsimulations are current state-of-the-art approaches. However, existingagent-based models still have some limitations on behavioral realism andresource demand that limit their applicability. In this study, leveraging theemerging technology of large language models (LLMs) and LLM-based agents, wepropose a general LLM-agent-based modeling framework for transportationsystems. We argue that LLM agents not only possess the essential capabilitiesto function as agents but also offer promising solutions to overcome somelimitations of existing agent-based models. Our conceptual framework designclosely replicates the decision-making and interaction processes and traits ofhuman travelers within transportation networks, and we demonstrate that theproposed systems can meet critical behavioral criteria for decision-making andlearning behaviors using related studies and a demonstrative example of LLMagents' learning and adjustment in the bottleneck setting. Although furtherrefinement of the LLM-agent-based modeling framework is necessary, we believethat this approach has the potential to improve transportation system modelingand simulation.