Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

LLM-ABM for Transportation: Assessing the Potential of LLM Agents in System Analysis

Tianming Liu, Jirong Yang, Yafeng Yin

TL;DR

This paper addresses limitations in traditional agent-based transportation models, such as bounded rationality and calibration data needs, by integrating LLM-based traveler agents into an ABM framework. It combines memory, reflection, and bounded rationality prompts with a Dynamic Traffic Assignment (DTA) simulator to evaluate both individual departure/route decisions and resulting network dynamics in the morning commute. The authors demonstrate that LLM agents exhibit sound individual behavior and that the emergent system dynamics converge to established benchmarks, including arrival distributions and Wardrop equilibrium, underscoring the potential of LLMs for transportation planning. This work provides a path toward more flexible, data-efficient, and policy-sensitive planning tools, while outlining future extensions to multimodal networks and richer traveler personas.

Abstract

Agent-based modeling approaches represent the state-of-art in modeling travel demand and transportation system dynamics and are valuable tools for transportation planning. However, established agent-based approaches in transportation rely on multi-hierarchical mathematical models to simulate travel behavior, which faces theoretical and practical limitations. The advent of large language models (LLM) provides a new opportunity to refine agent-based modeling in transportation. LLM agents, which have impressive reasoning and planning abilities, can serve as a proxy of human travelers and be integrated into the modeling framework. However, despite evidence of their behavioral soundness, no existing studies have assessed the impact and validity of LLM-agent-based simulations from a system perspective in transportation. This paper aims to address this issue by designing and integrating LLM agents with human-traveler-like characteristics into a simulation of a transportation system and assessing its performance based on existing benchmarks. Using the classical transportation setting of the morning commute, we find that not only do the agents exhibit fine behavioral soundness, but also produce system dynamics that align well with standard benchmarks. Our analysis first verifies the effectiveness and potential of LLM-agent-based modeling for transportation planning on the system level.

LLM-ABM for Transportation: Assessing the Potential of LLM Agents in System Analysis

TL;DR

This paper addresses limitations in traditional agent-based transportation models, such as bounded rationality and calibration data needs, by integrating LLM-based traveler agents into an ABM framework. It combines memory, reflection, and bounded rationality prompts with a Dynamic Traffic Assignment (DTA) simulator to evaluate both individual departure/route decisions and resulting network dynamics in the morning commute. The authors demonstrate that LLM agents exhibit sound individual behavior and that the emergent system dynamics converge to established benchmarks, including arrival distributions and Wardrop equilibrium, underscoring the potential of LLMs for transportation planning. This work provides a path toward more flexible, data-efficient, and policy-sensitive planning tools, while outlining future extensions to multimodal networks and richer traveler personas.

Abstract

Agent-based modeling approaches represent the state-of-art in modeling travel demand and transportation system dynamics and are valuable tools for transportation planning. However, established agent-based approaches in transportation rely on multi-hierarchical mathematical models to simulate travel behavior, which faces theoretical and practical limitations. The advent of large language models (LLM) provides a new opportunity to refine agent-based modeling in transportation. LLM agents, which have impressive reasoning and planning abilities, can serve as a proxy of human travelers and be integrated into the modeling framework. However, despite evidence of their behavioral soundness, no existing studies have assessed the impact and validity of LLM-agent-based simulations from a system perspective in transportation. This paper aims to address this issue by designing and integrating LLM agents with human-traveler-like characteristics into a simulation of a transportation system and assessing its performance based on existing benchmarks. Using the classical transportation setting of the morning commute, we find that not only do the agents exhibit fine behavioral soundness, but also produce system dynamics that align well with standard benchmarks. Our analysis first verifies the effectiveness and potential of LLM-agent-based modeling for transportation planning on the system level.

Paper Structure

This paper contains 18 sections, 4 equations, 5 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related work
  3. Agent-based Models in Transportation
  4. LLM Agents in Human-Behavior Simulations
  5. Framework
  6. Preliminaries
  7. Agent behavior
  8. Memory
  9. Reflection
  10. Traffic Flow Simulation
  11. Experiments
  12. Continuous choice case: commute departure time choice
  13. Setup
  14. Result and analysis
  15. Discrete choice case: commute route choice
  16. ...and 3 more sections

Figures (5)

  • Figure 1: Illustration of the LLM agent perception and decision-making pipeline
  • Figure 2: Behavioral pipeline of LLM traveler agents
  • Figure 3: Distribution of Arrival Times: First Day(Group) vs. Last Day(Group)
  • Figure 4: Distribution of Departure Times: First 5 Days vs. Last 5 Days
  • Figure 5: Progression of number of agents on each route