Extract component models using introspection at runtime []

This tutorial requires a local ROS installation.

For the auto-generation of models another strategy, parallel to static code analysis techniques is the introspection of a running system.

The tool ros2model was created for that purpose. It can be installed as every common ROS 2 package by:

1. clone this repository into the source folder in your workspace, such as:

   ws/src/ros2model
   

2. Source ROS workspace

   source /opt/ros/humble/setup.bash
   

3. Back to the folder “ws”, create and active venv

   python3 -m venv venv --system-site-packages --symlinks
   source venv/bin/activate
   

4. Install poetry and install dependencies run

   pip install poetry
   poetry install -C src/ros2model/
   

5. compile it as ROS package

   python -m colcon build --packages-select ros2model --symlink-install
   

   source install/local_setup.bash
   

6. config python path in this case we are using python3.10 in venv

   export PYTHONPATH=$PYTHONPATH:$(pwd)/venv/lib/python3.10/site-packages/
   

Then you can start the nodes you want to analyze. For example the common turtlesim tutorial:

ros2 launch turtlesim multisim.launch.py

And then just call the extractor to generate the models:

ros2 model node -o turtlesim/nodes

This command will create automatically under the selected folder (in this case turtlesim/nodes) the .ros2 files corresponding to all the running nodes.

You can copy the files to the ROS project in Eclipse to verify the models.

Please note that some information can be wrong as from the runtime system we can not get the filesystem information like the name of the package and the name of the artifacts. This information must be reviewed by the user.

Many of the models you will see during the rest of the tutorials were auto-generated using this method.