Starling is an agent-based simulation framework for urban mobility.
It provides generic classes to build transport models such as station-based sharing system, public transport or shared taxis.
Citation info: Leblond, V, Desbureaux, L, Bielecki, V., 2020. "A new agent-based software for designing and optimizing emerging mobility services : application to city of Rennes." Proceedings of the European Transport Conference, 11 September 2020, Milan
Feel free to use and contribute to Starling project as long as you comply with the licence terms.
This section will show you how to run the example simulation scenarios.
For a more detailed setup guide, see the section Running simulations of the documentation.
Starling must be cloned locally in order to be run.
git clone https://github.com/tellae/starling.gitThen, you can either install the dependencies directly on your linux or use a Docker container to run a simulation.
We recommend the Linux installation for development and the Docker installation for running simulations.
This procedure is described for a Linux Ubuntu 18.04 or 20.04 with Python 3.6 or higher already installed.
First, install the necessary Linux packages with
sudo apt-get install -yy -q libcurl4-gnutls-dev \
libssl-dev libproj-dev libgdal-dev gdal-bin python3-gdal \
libgdal-dev libudunits2-dev pkg-config libnlopt-dev libxml2-dev \
libcairo2-dev libudunits2-dev \
libgdal-dev libgeos-dev libproj-dev python3-pip python3-dev \
build-essential libspatialindex-dev python3-rtreeThen, install the Python libraries using pip3
# upgrade pip
python3 -m pip install --upgrade pip
# install the project requirements
pip3 install -r requirements.txtRun the following command to create a Docker image named starling containing python and all requirements. This image doesn’t contain Starling source code but it contains all python dependencies for running Starling.
docker build . --tag="starling"You can run Docker in interactive mode (which will place you inside the container, as in a terminal) with the following command:
Linux
docker run -it -v "$(pwd):/starling_dir/" --name container_name starlingWindows
docker run -it -v "%cd%:/starling_dir/" --name container_name starlingDocker can also be run in detached mode, which lets the simulations run on their own (see the documentation).
You can now build the data structure and download example scenarios by running the following command in your environment
python3 main.py -eOnce the data is prepared, a scenario can be run from the project root by running main.py with the path to the scenario folder.
Run one of the example scenarios, for instance:
python3 main.py data/models/SB_VS/example_nantes/You will see the progression of the simulation with the logs that appear in the console.
You can find the outputs of the scenario in the output folder. In this case, its data/models/SB_VS/example_nantes/outputs/.
KPI files (.csv.gz) can be visualised with any spreadsheet software.
The visualisation file (.geojson) can be uploaded to the web application Kite to visualise the simulation run.
The project documentation is generated using Sphinx and hosted by Read the Docs here:
https://starling.readthedocs.io/en/latest/
The documentation is still incomplete, but contributions and suggestions are very welcome.
Feedback and contributions on Starling, the code, the documentation or even the github management are very welcome.
We haven't established a contributing procedure yet, but we will do our best to guide you if you want to contribute to this project.
You can email us at [email protected] for any support demand.
Tellae is actively contributing to Starling as part of its research and development activities.
Tellae uses Starling in commercial contracts to study mobility projects. This is why some of the models and algorithms developed around Starling are not shared in this repository.