Right now it seems like there are HRIM models for robot sensors only. In traditional network monitoring the state of network components like managed industrial switches, industrial wlan access points, etc. is gathered via protocols like SNMP or via other interfaces. What’s the state of the HRIM models for communication components?
Hello @fkromer,
Right now our main contributions to the HRIM project is in sensors and actuators. But we are really excited about adding more and more models to HRIM. Communication is a very interesting. Can you provide more details about what are you looking for?
We can start a discussion about what you need and introduce these new models into the HRIM project.
1 Like
Hi @ahcorde,
probably it’s necessary to add some additional contextual information to help people understand why there is the need for a HRIM model for Ethernet switches.
Motivation
Ethernet switches are potentially contained in ROS2 communication network infrastructures connecting several components of an overall robotics application.
The status of switches may lead to data throughput (safety) and security issues in the application. Monitoring the status of Ethernet switches is possible using external tools of the traditional IT monitoring domain. However integration of the observed data into the ROS2 application is not straightforward and the observed data may not be considered real-time.
A HRIM model for Ethernet switches would enable the integration of observed data into ROS2 application with better real-time characteristic.
Proposal
Ethernet switches are usually managed. This means they provide at least one interface which allows the configuration and observation of the switch configuration and status. Typical interfaces are protocols like SNMP, syslog, etc. .
A typical industrial Ethernet switch is e.g. the Cisco Industrial Ethernet 3000 Series Switches (datasheet). MIB files supported by the switch define what data may configured/observed. The HRIM model should reflect all characteristics of switches relevant for ROS2 applications.
I’ll have a look into the HRIM models. Probably I’ll think about and propose a model as PR.
3 Likes
Hi @fkromer,
Nice discussion! As @ahcorde explained we focused mainly in sensors and actuators so the communication part is completly open to discussion.
Here are some ideas that in my opinion would be the way to go for HRIM communication models.
I’ll have a look into the [HRIM models (HRIM/models at master · AcutronicRobotics/HRIM · GitHub). Probably I’ll think about and propose a model as PR.
I think in this what make makes more sense to me is to reuse the existing network management models (YANG models). A first approach could be a simple wrapping of these models in HRIM.
The status of switches may lead to data throughput (safety) and security issues in the application. Monitoring the status of Ethernet switches is possible using external tools of the traditional IT monitoring domain. However integration of the observed data into the ROS2 application is not straightforward and the observed data may not be considered real-time.
For this topic the OMG DDS-TSN proposal is going to be very relevant. One of the aims of this proposal is to be able to configure a TSN network based on the DDS QoS requirements. I think this fits very well with the aim of HRIM because the network configuration would follow a more DDS approach, and hence, a more ROS 2 approach. For instance, a robotic component would specify their network requirements (bandwidth, deadline, LatencyBudget) and a network manager would try to configure the network in order to meet these requirements. If the network configuration fails (not enough bandwidth for example), the network manager would inform the component about the failure the ROS 2 application won’t reach an active status.
4 Likes