PyTrees ROS Viewer

[About] [Features] [Quickstart] [Usage] [Modding]

About

A Qt-JS application for visualisation of behaviour trees in a ROS 2 ecosystem. This implementation serves as the GUI visualisation tool for PyTrees ROS.

Features

• Tree Visualisation
• Collapsible Subtrees
• Zoom and Scale Contents to Fit
• Timeline Rewind & Resume
• Blackboard View
• Activity View

Quickstart

sudo apt install ros-<rosdistro>-py-trees-ros-tutorials
sudo apt install ros-<rosdistro>-py-trees-ros-viewer

# In a first shell
py-trees-tree-viewer
# In a second shell
ros2 launch py_trees_ros_tutorials tutorial_eight_dynamic_application_loading_launch.py
# Click 'Scan' on the qt robot dashboard interface
# Wear a colander, I am too.

Usage

Until a snapshot stream is discovered or selected, you’ll land at the splash screen which enumerates the interactive options available.

The viewer will continuously scan for and update the namespace drop-down with a list of py_tree instances (specifically, the namespace of their snapshot stream services). If not yet connected, it will provide the convenience of automatically making a connection to the first instance it discovers. Beyond that, it is possible to switch between streaming services via the drop-down.

At a minimum, the stream will send updates to the tree graph when the tree changes (i.e. when any one of the behaviours modifies it’s status). Additional configuration of the stream can be managed via the checkboxes in the top right of the window - introspect on the blackboard and/or request more frequent updates (useful when tracking blackboard changes).

If you have a large tree, collapse sections of it. This is useful in tandem with the screenshot capability when you wish to highlight an area of interest to communicate a design problem or report a bug.

In the same vein, filter the blackboard to those keys expicitly associated (used) by manually selected behaviours.

Problems debugging a catastrophe and the final state of the tree doesn’t uncover the root cause, merely the mayhem that resulted? This is a very common situation - use the timeline! Rewind to a previous state and time travel. Note: the stream configuration only applies to future messages - you can’t change the past! If you are concerned about being able to debug affectively, make sure you have at least the ‘Blackboard Data’ and ‘Periodic’ checkboxes enabled so that time travelling along the timeline has sufficient data for analysis.

Travelling along the timeline is possible even while online. Simply hit the ‘Resume’ button to resume visualisation of the live feed.

Modding

The underlying py_trees_js library lends itself to being used in various ways. This application can be used as a baseline reference (since it utilises the underlying py_trees_js library fully) for modifications to suit your own use case. Open an issue if you have questions and/or would like assistance.

1) Visualisation for a behaviour trees implementation that is not py_trees.

Replicate the hybrid qt-js application here and replace backend.py. The data structure eventually passed to the js library is a json dictionary that only makes use of the most fundamental properties of behaviour trees (e.g. tree graph structure, behaviour status) and consequently, it should be a matter of merely wiring connections and converting data.

2) Visualisation for a different middleware architecture

File truncated at 100 lines see the full file