Contributions are welcome, and they are greatly appreciated! Every little bit helps, and credit will always be given.

You can contribute in many ways:

Types of Contributions

Report Bugs

Report bugs at

If you are reporting a bug, please include:

  • Your operating system name and version.
  • Any details about your local setup that might be helpful in troubleshooting.
  • Detailed steps to reproduce the bug.

Fix Bugs

Look through the Repository Issue Tracker for bugs. Anything tagged with “bug” and “help wanted” is open to whoever wants to implement it.

Implement Features

Look through the Repository Issue Tracker for features. Anything tagged with “enhancement” and “help wanted” is open to whoever wants to implement it.

Write Documentation

flatland could always use more documentation, whether as part of the official flatland docs, in docstrings, or even on the web in blog posts, articles, and such. A quick reference for writing good docstrings is available at :

Submit Feedback

The best way to send feedback is to file an issue at

If you are proposing a feature:

  • Explain in detail how it would work.
  • Keep the scope as narrow as possible, to make it easier to implement.
  • Remember that this is a volunteer-driven project, and that contributions are welcome :)

Get Started!

Ready to contribute? Here’s how to set up flatland for local development.

  1. Fork the flatland repo on .

  2. Clone your fork locally:

    $ git clone
  3. Install the software dependencies via Anaconda-3 or Miniconda-3. (This assumes you have Anaconda installed by following the instructions here)

    $ conda install -c conda-forge tox-conda $ conda install tox $ tox -v –recreate

    This will create a virtual env you can then use.

    These steps are performed if you run

    $ getting_started/getting_started.bat/.sh

    from Anaconda prompt.

  4. Create a branch for local development:

    $ git checkout -b name-of-your-bugfix-or-feature

    Now you can make your changes locally.

  5. When you’re done making changes, check that your changes pass flake8 and the tests, including testing other Python versions with tox:

    $ flake8 flatland tests examples benchmarks
    $ python test or py.test
    $ tox

    To get flake8 and tox, just pip install them into your virtualenv.

  6. Commit your changes and push your branch to Gitlab:

    $ git add .
    $ git commit -m "Addresses #<issue-number> Your detailed description of your changes."
    $ git push origin name-of-your-bugfix-or-feature
  7. Submit a merge request through the Gitlab repository website.

Merge Request Guidelines

Before you submit a merge request, check that it meets these guidelines:

  1. The merge request should include tests.
  2. The code must be formatted (PyCharm)
  3. If the merge request adds functionality, the docs should be updated. Put your new functionality into a function with a docstring, and add the feature to the list in README.rst.
  4. The merge request should work for Python 3.6, 3.7 and for PyPy. Check and make sure that the tests pass for all supported Python versions. We force pipelines to be run successfully for merge requests to be merged.
  5. Although we cannot enforce it technically, we ask for merge requests to be reviewed by at least one core member in order to ensure that the Technical Guidelines below are respected and that the code is well tested:

5.1. The remarks from the review should be resolved/implemented and communicated using the ‘discussions resolved’:


5.2. When a merge request is merged, source branches should be deleted and commits squashed:



To run a subset of tests:

$ py.test tests.test_flatland


A reminder for the maintainers on how to deploy. Make sure all your changes are committed . Then run:

$ bumpversion patch # possible: major / minor / patch
$ git push
$ git push --tags

TODO: Travis will then deploy to PyPI if tests pass. (To be configured properly by Mohanty)

Local Evaluation

This document explains you how to locally evaluate your submissions before making an official submission to the competition.


  • flatland-rl : We expect that you have flatland-rl installed by following the instructions in [](
  • redis : Additionally you will also need to have redis installed and should have it running in the background.

Test Data

  • test env data : You can download and untar the test-env-data <>, at a location of your choice, lets say /path/to/test-env-data/. After untarring the folder, the folder structure should look something like:
└── test-env-data
    ├── Test_0
    │   ├── Level_0.pkl
    │   └── Level_1.pkl
    ├── Test_1
    │   ├── Level_0.pkl
    │   └── Level_1.pkl
    ├── Test_8
    │   ├── Level_0.pkl
    │   └── Level_1.pkl
    └── Test_9
        ├── Level_0.pkl
        └── Level_1.pkl

Evaluation Service

  • start evaluation service : Then you can start the evaluator by running :
flatland-evaluator --tests /path/to/test-env-data/


  • run client : Some sample submission code can be found in the starter-kit, but before you can run your code locally using FlatlandRemoteClient, you will have to set the AICROWD_TESTS_FOLDER environment variable to the location where you previous untarred the folder with the test-env-data:
export AICROWD_TESTS_FOLDER="/path/to/test-env-data/"

# or on Windows :
# set AICROWD_TESTS_FOLDER "\path\to\test-env-data\"

# and then finally run your code

Technical Guidelines

Clean Code

Please adhere to the general Clean Code principles, for instance we write short and concise functions and use appropriate naming to ensure readability.

Naming Conventions

We use the pylint naming conventions:

module_name, package_name, ClassName, method_name, ExceptionName, function_name, GLOBAL_CONSTANT_NAME, global_var_name, instance_var_name, function_parameter_name, local_var_name.


Docstrings should be formatted using numpydoc.

Acessing resources

We use importlib-resources to read from local files.

Sample usages:

from importlib_resources import path

with path(package, resource) as file_in:
    new_grid = np.load(file_in)


from importlib_resources import read_binary

load_data = read_binary(package, resource)

Renders the scene into a image (screenshot)"filename.bmp")

Type Hints

We use Type Hints (PEP 484) for better readability and better IDE support.

Have a look at the Type Hints Cheat Sheet to get started with Type Hints.

Caveat: We discourage the usage of Type Aliases for structured data since its members remain unnamed (see Issue #284).


For structured data containers for which we do not write additional methods, we use NamedTuple instead of plain Dict to ensure better readability by

Members of NamedTuple can then be accessed through .<member> instead of [‘<key>’].

If we have to ensure some (class) invariant over multiple members (for instance, o.A always changes at the same time as o.B), then we should uses classes instead, see the next section.

Class Attributes

We use classes for data structures if we need to write methods that ensure (class) invariants over multiple members, for instance, o.A always changes at the same time as o.B. We use the attrs class decorator and a way to declaratively define the attributes on that class:

Abstract Base Classes

We use the abc class decorator and a way to declaratively define the attributes on that class:

And then


import abc
from abc_base import PluginBase

class SubclassImplementation(PluginBase):

    def load(self, input):

    def save(self, output, data):
        return output.write(data)

if __name__ == '__main__':
    print('Subclass:', issubclass(SubclassImplementation,
    print('Instance:', isinstance(SubclassImplementation(),


We discourage currying to encapsulate state since we often want the stateful object to have multiple methods (but the curried function has only its signature and abusing params to switch behaviour is not very readable).

Thus, we should refactor our generators and use classes instead (see Issue #283).