AgentSUMO: An Agentic Framework for Interactive Simulation Scenario Generation in SUMO via Large Language Models

TL;DR

AgentSUMO addresses the barrier that non-experts face when using SUMO by replacing imperative, command-driven workflows with an agentic, goal-oriented approach. It introduces an Interactive Planning Protocol for adaptive reasoning and a Model Context Protocol to standardize tool orchestration, enabling natural-language-driven scenario generation, policy experimentation, and result analysis. Across Seoul and Manhattan, the framework autonomously derives executable SUMO workflows from abstract objectives, performs multi-step planning, and yields interpretable, policy-relevant insights that are reproducible and scalable. This work advances urban mobility decision-support by marrying LLM-driven reasoning with protocol-based tool integration to bridge high-level aims and detailed simulation execution.

Abstract

The growing complexity of urban mobility systems has made traffic simulation indispensable for evidence-based transportation planning and policy evaluation. However, despite the analytical capabilities of platforms such as the Simulation of Urban MObility (SUMO), their application remains largely confined to domain experts. Developing realistic simulation scenarios requires expertise in network construction, origin-destination modeling, and parameter configuration for policy experimentation, creating substantial barriers for non-expert users such as policymakers, urban planners, and city officials. Moreover, the requests expressed by these users are often incomplete and abstract-typically articulated as high-level objectives, which are not well aligned with the imperative, sequential workflows employed in existing language-model-based simulation frameworks. To address these challenges, this study proposes AgentSUMO, an agentic framework for interactive simulation scenario generation via large language models. AgentSUMO departs from imperative, command-driven execution by introducing an adaptive reasoning layer that interprets user intents, assesses task complexity, infers missing parameters, and formulates executable simulation plans. The framework is structured around two complementary components, the Interactive Planning Protocol, which governs reasoning and user interaction, and the Model Context Protocol, which manages standardized communication and orchestration among simulation tools. Through this design, AgentSUMO converts abstract policy objectives into executable simulation scenarios. Experiments on urban networks in Seoul and Manhattan demonstrate that the agentic workflow achieves substantial improvements in traffic flow metrics while maintaining accessibility for non-expert users, successfully bridging the gap between policy goals and executable simulation workflows.

Figures (9)

• Figure 1: Illustration of the Motivation for AgentSUMO. The framework enables non-expert users to perform expert-level traffic simulations by converting high-level goals (e.g., “reduce congestion”) into executable SUMO workflows without manual scripting.
• Figure 2: Overall framework of AgentSUMO.
• Figure 3: Qualitative results of Simple Tasks across two regions under varying traffic conditions.
• Figure 4: llustration of complex task reasoning and execution pipeline in AgentSUMO (lane-closure and EV-adoption scenarios)
• Figure 5: Interactive scenario generation and executed outcome for the lane-closure scenario on Teheran-ro.