TL;DR
Support the development of a novel ROS 2 visualization approach by sharing your ROS 2 system as dataset here: vschroeter/rosmetasys-datasets (github.com)
Dear ROS Community,
At the Hasso Plattner Institute, Digital Engineering Faculty, University of Potsdam, we are currently researching new visualization approaches for communication graphs.
Such approaches could be alternatives to the currently used rqt_graph.
We do not want to visualize the data transferred between nodes like RViz etc., but instead the nodes of the system on an abstract level. So, the insights that the visualization should provide are:
- What nodes exist in the system?
- How are the nodes connected?
- Which nodes are unconnected / strongly connected?
- What global and local data flows are indicated by the connections between the nodes?
- What channels (topics for Publishers/Subscribers and Services) are offered by a node?
- Showing live to changes in the system (new nodes, stopped nodes, changed topics, etc.)
We are currently testing an algorithm suite on datasets with only about 30 to 40 nodes. To ensure, that the approach is scalable for and applicable to larger, real-world systems, we would be very grateful if you could support our development and evaluation of the algorithms by sharing your own datasets (anonymized if desired).
Steps to export your dataset:
- Install rosmetasys with pip
pip install rosmetasys. - Start your ROS system.
- Take a snapshot of your system by running
rosmetasys export -i.
To share your dataset:
- Either submit a new issue with your dataset attached.
- Or fork the rosmetasys-datasets repository, add your dataset, and create a new pull request (this automatically makes you a contributor to the repository).
- Or e-Mail us the dataset to valentin.schroeter@hpi.uni-potsdam.de (if you don’t want your dataset to be public, but still want to support our development).
We appreciate any dataset, regardless of the node count, structure, etc. Especially large datasets would be interesting to ensure the quality of our approach.
You can also visualize your dataset (or see sample data) in a working prototype, which we have published as an intermediate step of our work:
- Open the website and upload the dataset using the
Upload JSON fileinput field in the left drawer. - Take a look at the circular overview.
- View the details of a node by double-clicking it in the overview.
Thanks a lot in advance for every contribution!
