Traffic Simulation with SUMO: A Developer's Handbook


Traffic Simulation with SUMO: A Developer's Handbook

eclipse-sumo/sumo

2025-08-10

From a software engineer's perspective, SUMO is more than just a traffic simulator—it's a robust platform for testing and developing applications related to transportation, logistics, and smart cities.

Algorithm Testing
Do you have a new algorithm for optimizing traffic light timing, route planning, or autonomous vehicle (AV) control? SUMO provides a realistic, virtual environment to test your code without needing to deploy it on a real-world network, which can be expensive and dangerous.

Data Generation
SUMO can generate vast amounts of realistic traffic data, including vehicle positions, speeds, and routes. This data is invaluable for training machine learning models, for example, to predict traffic flow or detect anomalies.

Integration with Other Systems
SUMO can be integrated with other software platforms. You can create a system where your code controls the vehicles or traffic lights within the simulation, giving you a full-stack environment for developing and testing complex systems.

Scenario Modeling
The package allows you to model complex scenarios, like the impact of a new road, the effect of an accident, or the optimal placement of charging stations for electric vehicles. This is crucial for developing tools for urban planning and resource management.

Getting started with SUMO is straightforward. The project is open-source and well-documented.

You can install SUMO on your system using various methods.

Package Manager
On Linux, you can often install it with your system's package manager. For example, on Ubuntu, you would use sudo apt-get install sumo.

From Source
You can also download the source code from the official GitHub repository (eclipse-sumo/sumo) and compile it yourself. This gives you more control over the build process.

Pre-compiled Binaries
The SUMO website provides pre-compiled binaries for different operating systems (Windows, macOS, Linux), which is the easiest way to get started.

A basic SUMO simulation requires a few key files

Network file (.net.xml)
This file defines the roads, junctions, and lanes. You can either create this manually or, more commonly, import a real-world network from OpenStreetMap using the provided netconvert tool.

Routes file (.rou.xml)
This file defines the vehicles, their routes, and their departure times. You can either define specific routes or use a tool like randomTrips.py to generate random traffic.

Configuration file (.sumo.cfg)
This file ties everything together, pointing to the network and routes files and setting other simulation parameters.

Once you have these files, you can run the simulation from the command line

sumo-gui -c your_config.sumo.cfg

The -gui option opens the graphical user interface, allowing you to visualize the traffic flow.

SUMO has a very useful feature called TraCI (Traffic Control Interface), which allows you to control and retrieve data from a running simulation in real-time. This is where the real power for a software engineer lies. You can use languages like Python to control the simulation.

Here's a simple Python script using traci to change a traffic light's state.

First, you need to start SUMO in client mode

sumo --remote-port 8813 --step-length 0.1 -c your_config.sumo.cfg

Then, in a separate terminal, you can run your Python script

import traci
import time

# Connect to the SUMO simulation
traci.start(["sumo-gui", "-c", "your_config.sumo.cfg"])

# Run the simulation for a few steps
step = 0
while step < 1000:
    traci.simulationStep()

    # Let's say we want to change the traffic light "gneJ7" to red for a few steps
    if step == 50:
        print("Changing traffic light to red!")
        # The 'r' here stands for red
        traci.trafficlight.setRedYellowGreenState("gneJ7", "rrrr")

    # Change it back to green
    if step == 100:
        print("Changing traffic light back to green!")
        # The 'G' here stands for green
        traci.trafficlight.setRedYellowGreenState("gneJ7", "GGGG")
    
    step += 1

# Close the connection
traci.close()

print("Simulation finished.")

This example demonstrates how you can programmatically interact with the simulation. You can use this same approach to develop more complex logic, like dynamic routing based on real-time traffic, or to control the behavior of individual vehicles.


eclipse-sumo/sumo