sumo_docker_pipeline

The package sumo_docker_pipeline enables you to run a traffic simulator SUMO efficiently and to interact with Python easily. The package is valid when you need to run SUMO iteratively.

SUMO is often tricky to install locally because of its dependencies. Thus, it's a straightforward idea to run SUMO in a docker container.

However, another issue arises when we run SUMO in a docker container. It is challenging to construct a pipeline between SUMO and API.

The package provides an easy-to-use API; at the same time, SUMO runs in a docker container.

Requirement

• python > 3.5
• docker
• docker-compose

Install

Pull the image (or build of a docker image with SUMO)

The existing image is on the Dockerhub.

docker pull kensukemi/sumo-ubuntu18

If you prefer to build with yourself, you run the following command.

docker-compose build 

Install a python package

make install

Example case: iterative run with parameter updates

Let's say that you want to run SUMO iteratively. At the same time, you want to change input parameters depending on the results of a simulation.

In that case, you need to check the output of SUMO and update the parameters. sumo_docker_pipeline package enables you to make the process automatic.

Setups

1. creation of a directory where you save SUMO's configuration.
2. creation of template-files of SUMO's configuration.
3. running the pipeline.

1. creation of a directory

It is a directory that SUMO accesses. Let's say that we create test/resources/config_template

2. creation of SUMO's configuration

You prepare configuration files which SUMO requires. The format of the conf. files are totally same as SUMO's requirements.

The only difference is that you write wildcard ? at the place where you wanna replace during pipeline.

For example, tests/resources/config_template/grid.flows.xml has the following element,

<vType vClass="passenger" id="passenger"  tau="0.5" speedDev="0.1" maxSpeed="?" minGap="?" accel="?" decel="?" latAlignment="center" />

With the sumo_docker_pipeline package, you can replace the attributes with the wildcards ?.

3. running the pipeline

See examples directory.

For developer

pytest tests

license and credit

The source code is licensed MIT. The website content is licensed CC BY 4.0.

@misc{sumo-docker-pipeline,
  author = {Kensuke Mitsuzawa},
  title = {sumo-docker-pipeline},
  year = {2021},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/Kensuke-Mitsuzawa/sumo_docker_pipeline}},
}