Self-organizing traffic lights in SUMO

⚠️ I'm not working on this tools anymore, they may or may not work since I last used them in 2018.

Description

This project contains some tools for the open source traffic simulator SUMO that I used for my masters degree that may be of some use for some one else. The main purpose of this files was to create some networks to test the difference between self-organizing traffic lights and other traffic light models.

Overview

To create a new simulation with this files, you need 7 files:

1. Network
2. Detectors
3. Configuration
4. Output
5. Traffic light programs
6. Routes

Once all the files are created you can run a simulation with the runner.py script.

❗ All the scripts mentioned below have a -h flag to have more details about the options.

Network

To create the network files first you need to take a map from Open Street Maps. Then edit it with JOSM until it looks as you want in SUMO. To test it in SUMO run the newMap.py script inside the newmap folder and run the runner.py script. newMap.py creates all the basic files (except the routes) but there's not going to be any simulation yet.

Routes

To generate some routes for the cars you can use the generateRoutes.py script inside the input folder. The basic command is:

python generateRoutes.py -n map -p 0.1 -c 1

Which generates 0.1% of all routes with 1 vehicle per route for the map called map.

Simulation

Once the routes are added you can run a new simulation with the main runner.py script. For example, create an output folder in output/map. Then, to run a simulation with the self-organizing method (default) using the map created with the previous scripts and with sumo-gui to visualize it you use:

python runner.py --dirNet map -g

In this case the script uses the files inside the input/map folder, if they don't exist it will fail. The output folder must also be created before running the script.

Results

If everything works as it should you should get results inside the output/map folder.

If you are using the self-organizing traffic lights you can analyze the results with the parseResults.py script inside the output folder. To run it you should run something similar to:

python parseResults.py -d map -s 0 -g

This should produce some .txt files containing the shanon information for every street, traffic light and intersection with a traffic light in the simulation. There's also files with the occupancy, mean speed and mean waiting time of the streets divided by priority. All this files produce a graphs inside the graphs/map folder. This graphs are made with the graphMaker.R script inside the graphs folder.