Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

TrafficSimAgent: A Hierarchical Agent Framework for Autonomous Traffic Simulation with MCP Control

Yuwei Du, Jun Zhang, Jie Feng, Zhicheng Liu, Jian Yuan, Yong Li

TL;DR

TrafficSimAgent presents a hierarchical, LLM-driven multi-agent framework for autonomous traffic simulation that translates natural-language instructions into executable experiments via MCP-enabled tools. It combines robust semantic instruction understanding, dynamic task planning, and element-level embodiment with a two-layer optimization system that mixes high-level strategy selection and low-level real-time collaboration. The approach demonstrates superior generality across online/offline scenarios and achieves competitive or superior optimization performance compared to baselines, including traditional TSC algorithms and other LLM-based methods. By decoupling API calls through MCP and enabling cross-agent collaboration and memory-informed reflection, the framework lowers barriers to complex traffic experiments and supports iterative, autonomous improvement with practical implications for transportation planning and policy testing.

Abstract

Traffic simulation is important for transportation optimization and policy making. While existing simulators such as SUMO and MATSim offer fully-featured platforms and utilities, users without too much knowledge about these platforms often face significant challenges when conducting experiments from scratch and applying them to their daily work. To solve this challenge, we propose TrafficSimAgent, an LLM-based agent framework that serves as an expert in experiment design and decision optimization for general-purpose traffic simulation tasks. The framework facilitates execution through cross-level collaboration among expert agents: high-level expert agents comprehend natural language instructions with high flexibility, plan the overall experiment workflow, and invoke corresponding MCP-compatible tools on demand; meanwhile, low-level expert agents select optimal action plans for fundamental elements based on real-time traffic conditions. Extensive experiments across multiple scenarios show that TrafficSimAgent effectively executes simulations under various conditions and consistently produces reasonable outcomes even when user instructions are ambiguous. Besides, the carefully designed expert-level autonomous decision-driven optimization in TrafficSimAgent yields superior performance when compared with other systems and SOTA LLM based methods.

TrafficSimAgent: A Hierarchical Agent Framework for Autonomous Traffic Simulation with MCP Control

TL;DR

TrafficSimAgent presents a hierarchical, LLM-driven multi-agent framework for autonomous traffic simulation that translates natural-language instructions into executable experiments via MCP-enabled tools. It combines robust semantic instruction understanding, dynamic task planning, and element-level embodiment with a two-layer optimization system that mixes high-level strategy selection and low-level real-time collaboration. The approach demonstrates superior generality across online/offline scenarios and achieves competitive or superior optimization performance compared to baselines, including traditional TSC algorithms and other LLM-based methods. By decoupling API calls through MCP and enabling cross-agent collaboration and memory-informed reflection, the framework lowers barriers to complex traffic experiments and supports iterative, autonomous improvement with practical implications for transportation planning and policy testing.

Abstract

Traffic simulation is important for transportation optimization and policy making. While existing simulators such as SUMO and MATSim offer fully-featured platforms and utilities, users without too much knowledge about these platforms often face significant challenges when conducting experiments from scratch and applying them to their daily work. To solve this challenge, we propose TrafficSimAgent, an LLM-based agent framework that serves as an expert in experiment design and decision optimization for general-purpose traffic simulation tasks. The framework facilitates execution through cross-level collaboration among expert agents: high-level expert agents comprehend natural language instructions with high flexibility, plan the overall experiment workflow, and invoke corresponding MCP-compatible tools on demand; meanwhile, low-level expert agents select optimal action plans for fundamental elements based on real-time traffic conditions. Extensive experiments across multiple scenarios show that TrafficSimAgent effectively executes simulations under various conditions and consistently produces reasonable outcomes even when user instructions are ambiguous. Besides, the carefully designed expert-level autonomous decision-driven optimization in TrafficSimAgent yields superior performance when compared with other systems and SOTA LLM based methods.
Paper Structure (27 sections, 5 figures, 7 tables)

This paper contains 27 sections, 5 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Methods
  3. Overall Framework
  4. MCP client
  5. MCP Server
  6. Design for Enhanced Generality
  7. Robust Instruction Understanding through Semantic Comprehension
  8. Dynamic Task Planning through Hierarchical Decomposition
  9. Diverse Task Support through Element-Agent Embodiment
  10. Design for Autonomous Optimization
  11. High-Level Optimization Strategy Selection
  12. Low-Level Real-Time Collaborative Optimization
  13. Experiments
  14. Experimental Settings
  15. Baseline Models and Methods
  16. ...and 12 more sections

Figures (5)

  • Figure 1: The whole framework of TrafficSimAgent .
  • Figure 2: Case study example
  • Figure 3: Gender and age distribution of user data generated by TrafficSimAgent for different user groups.
  • Figure 4: Education distribution of user data generated by TrafficSimAgent for different user groups.
  • Figure 5: Metrics variation with simulation steps of different optimization methods.