Eclipse Muto provides an adaptive framework and a runtime platform for dynamically composable model-driven ROS software stacks. Eclipse Muto can be used to introspect, monitor and manipulate the actively running ROS graph (the network of ROS nodes).
Agent is a runtime ROS component (Node) acting as a transceiver between the edge device and the twin server (Eclipse Ditto). Agent acts as a gateway between the device and remote management capabilities with support for asynchronous communication capabilities via MQTT. It acts as a communication bridge between edge devices and their respective virtual twins. The main aspects of data transported by Agent are as follows:
Agent is capable of publishing data streams up to the Twin Server, such as data that streams in ROS topics. Stream of data mapped from edge device to virtual one, could represent instant telemetry information and/or any other device specific details. This kind of data is useful for monitoring devices and algorithmic parameters with adjustable-frequency updates. Data such as odometry, speed, localization, goal or any other drive and device related can be broadcasted via declarative models. This kind of fast paced, high frequency data moving up to digital twins is also read by Muto Dashboard to provide a graphical interface and an adequate representation of aforementioned data.
Agent relays commands from the twin server to Composer running on the edge device. This type of information may be related to the lifecycle of ROS nodes that constitute the software stack actively running on the edge device as well as the lifecycle actions (start, stop, update etc) that may trigger the composer to respond.
Composer is a runtime ROS component (Node) that is responsible for the life cycle of a stack; Composer uses various ROS packages and APIs to launch ROS Nodes defined by the software stacks. It can be thought of as a smart launch manager. Composer uses the stack information model and computes the delta between the current and the desired state of the ROS graph and manages the lifecycle of ROS nodes accordingly.
A simple extensible Web/Mobile application for centralized management of edge devices and software stacks. It provides a plugable micro front end architecture for extensibility and an exemplary view for managing edge devices and composable ROS stacks on these devices. It supports and uses Eclipse Ditto device twins technology (and therefore protocols such as MQTT and REST/HTTP) for managing ROS based devices. It is designed to be modular at architectural level to make it easy to extend for a multitude of ROS applications.
The dashboard provides the following functionalities:
- Node Viewer : A graphical visualization of the ROS computational network that represents the actively running ROS nodes of the current state. It displays the node-to-node and node-to-topic affiliations with introspection information ( name, subscriber, publisher etc.)
- Stack Panel
- List: An interface to list the stack models that describe the components of a software stack that complies with the information model stored in the twin server.
- Diff: An interface to describe the algebraic differences between two models in terms of nodes and node related parameters .
- Remote Control Actions: An interface to manage (start,stop, update) the lifecycle of the software stack on the edge device.
- ROS Controls: Dashboard provides a graphical UI for some of the common ROS cli commands.
Micro frontends is an architectural style where independently deliverable frontend applications are composed into a greater whole. Muto LiveUI has been developed for building any such extensible Web or Mobile apps, such as the Muto Dashboard.
We use/fork code and examples from open initiatives such as f1tenth.org Autonomous Vehicle System, a powerful and versatile open-source platform for autonomous systems research and education, and Autoware.org an open-source software stack for self-driving vehicles to demonstrate dynamically adaptive ROS stacks. These examples provide a common repository for learning and teaching ROS and adaptive AD capabilities.
The content of this open source project is received and distributed under the license(s) listed above. Some source code and binaries may be distributed under different terms. Specific license information is provided in file headers and in NOTICE files distributed with the project's binaries.