ROS computation graph management solution

edit

Hi everyone,

i´m a german software engineering student and new to ROS. At the moment i´m working on my bachleor thesis. The goal is to design and implement a ROS node management library to control and observe nodes.

The requirements are:

• request and control the internal state of all running nodes (start, initializing, stop, shutdown, reboot).
• request various information of the nodes, such as plausibility check, cycle times and HeartBeat / AliveMessages.
• request various information the host system, such as CPU usage, RAM usage.

All of those information and control functions needs to be displayed in a hierarchical view, comparable to the OSI-layer model. The lower level contains the hardware drivers, followed by the data-processing processes. The upper layers are the abstract systems, such as planning.

During my research i found the diagnostic system and the bond package, which covers some of the requirements.

Are any existing approaches in discussion or is the need for this solution totally new?

Thank you very much in advanced for your answer.

Best regards