Speak to Simulate: Making Traffic Simulation Easy with AI Agents

What is it about?

City planners and policymakers often need to explore “what if” ideas, such as how to ease rush hour congestion or improve traffic flow. Yet most traffic simulators depend on complicated scripts, large datasets, and technical skills that make them difficult for non-experts to use. Our work presents AgentSUMO, an AI agent that makes traffic simulation as easy as having a conversation. Instead of writing code, users can simply describe their goals in everyday language. Even when a request is abstract or incomplete like saying “find a way to reduce traffic around downtown”, the AI interprets the intention, asks follow-up questions, fills in the missing details, and automatically builds and runs a simulation in SUMO. The agent does more than follow commands. It works with the user to plan experiments, test policies such as signal coordination or road closures, and present the results in a clear and visual way. In experiments on a Manhattan road network, AgentSUMO showed that it can intelligently design and optimize traffic scenarios that improve travel time and traffic flow. In essence, AgentSUMO makes traffic simulation interactive and approachable, allowing people to explore and experiment with city mobility through simple conversation with AI.

Featured Image

Photo by Vishal Chokkala on Unsplash

Photo by Vishal Chokkala on Unsplash

Why is it important?

Traditional traffic simulators are powerful tools but remain difficult to use, limiting their value to technical experts. Our work introduces an AI-driven framework that makes simulation accessible to anyone through natural conversation. What makes this approach unique is that the AI agent does more than execute fixed commands. It interprets vague goals, asks clarifying questions, and interactively builds complete traffic scenarios. The timing is important as cities increasingly rely on data-driven tools to manage congestion and design sustainable mobility policies. By lowering the technical barrier, AgentSUMO enables planners, researchers, and policymakers to test and refine their ideas quickly before implementing them in the real world. This shift can make urban policy experimentation faster, more inclusive, and guided by stronger, simulation-based evidence.

Perspectives