The FloorPlan DSL is a model-driven approach to specify indoor environments. The language enables developers to model indoor environments by its components such as spaces, entryways, and other features such as windows and columns; allowing the re-creation of real-world environments, or the creation of new ones. The models can be transformed into 3D meshes and occupancy grid maps, which can be used to simulate robot navigation tasks in most robot simulators.
It is strongly recommended to use Docker to install and use the FloorPlan DSL and other tools.
If not installed already, install docker as documented here. From the root directory of this repository, build the docker image:
docker build . --tag floorplan:latest
This will build the container and assign the string floorplan:latest
as a tag to run the correct container. To run the container execute the following command after replacing the correct paths inside the angle brackets:
docker run -v $<local output folder>:/usr/src/app/output -v $<local input folder>:/usr/src/app/models -it floorplan:latest bash
Important required options of the command:
-v $<local output folder>:/usr/src/app/output
: This will map a local folder of the computer with the output folder of the container, this allows the FloorPlan DSL tooling to write the artefacts directly in the local computer.-v $<local input folder>:/usr/src/app/models
: This will map a local folder of the computer with the input folder of the container, any models in the local input folder can be accessed by the FloorPlan DSL tooling.-it
: required to make the container interactive.bash
: required to make the container interactive.Install all the requirements:
sudo apt-get install blender python3-pip python3-venv -y
First, create a virtual environment and activate it:
python -m venv .venv
source .venv/bin/activate
For Blender to regonize the virtual environment, add it to your PYTHONPATH
:
export PYTHONPATH=<Path to .venv directory>/lib/python3.11/site-packages
From the root directory of the repo, install the python packages by running:
pip install -e .
This will install the languages and generators. To confirm that the languages have been installed, execute the commands: textx list-languages
and textx list-generators
. The names of the languages should appear in the output:
> textx list-languages
...
floorplan-v1 (*.floorplan) floorplan-dsl[1.1.0] A language to model indoor environments
floorplan-v2 (*.fpm2) floorplan-dsl[1.1.0] A language to model floor plans (v2)
floorplan-variation (*.variation)floorplan-dsl[1.1.0] A language to variate models from ExSce
> textx list-generators
...
floorplan-v2 -> json-ld floorplan-dsl[1.1.0] Generate composable models in json-ld
floorplan-v1 -> floorplan-v2 floorplan-dsl[1.1.0] Convert from floorplan models from v1 to v2
floorplan-variation -> floorplan-v2floorplan-dsl[1.1.0] Generate variations of indoor environments from .floorplan models
[!WARNING] The generation of 3D meshes and occupancy grid maps is currently being moved to the scenery_builder repository. The instructions below may not work and/or may be outdated.
This tool is currently in active development. To use the tool you can execute the following command:
blender --background --python src/exsce_floorplan/exsce_floorplan.py --python-use-system-env -- <model_path>
Optionally, you can remove the --background
flag to see directly the result opened in Blender.
Note: The --
before <model_path>
is intentional and important.
An example model for a building is available here. To generate the 3D mesh and occupancy grid:
blender --background --python src/exsce_floorplan/exsce_floorplan.py --python-use-system-env -- models/examples/hospital.floorplan
That should generate the following files:
.
├── map
│ ├── hospital.pgm
│ └── hospital.yaml
└── mesh
└── hospital.stl
The output path for the generated models in configurable (see confg/setup.cfg and note they are relative paths from where you’re calling the command).
The .stl
mesh can now be used to specify the Gazebo models and included in a Gazebo world. See, for example, this tutorial.
To generate the JSON-LD representation of the FloorPlan model, simply use textX’s language generators:
textx generate <floorplan-v2 model> --target json-ld -o <output path>
For example:
textx generate models/examples/brsu_building_c_with_doorways.fpm2 --target json-ld -o .
The JSON-LD models will be generated using radians, the default internal unit for angles in the floor plan model. To generate the JSON-LD models using degrees, add --angle-unit deg
to your command.
Modelling an environment can be straightforward with some background information on how the concepts are specified and related to each other. This tutorial will explain the concepts of the language and how to position them in the environment. An overview of the concepts and their attributes is available here. A tutorial on the variation DSL is also available here.
If you use our tooling and approach please cite our paper as follows:
@inproceedings{parra2023iros,
author = {Parra, Samuel and Ortega, Argentina and Schneider, Sven and Hochgeschwender, Nico},
title = ,
booktitle = {Proceedings of the IEEE International Conference on Intelligent Systems and Robots (IROS)},
year = {2023}
}
This work is part of a project that has received funding from the European Union’s Horizon 2020 research and innovation programme SESAME under grant agreement No 101017258.