MicroROS multi client capacity

Hello!

I’m working on a project where I plan to use micro-ROS across 60 Teensy boards. Although the application is not strictly robotics-related, the ROS structure — especially topics and tf — fits the project’s architecture very well.

Each Teensy would publish a point cloud (up to 1500 points) and a corresponding tf topic. However, due to hardware limitations, I haven’t been able to test the setup in practice.

I couldn’t find solid information about how many micro-ROS clients a single agent can support, especially when dealing with moderate-size messages like point clouds and tf data.

Any insights or experience on scalability, performance limits, or potential bottlenecks with many clients would be greatly appreciated.

Thanks in advance!

Have you found any evidence that the agent can handle more than 1 client? I always thought about it as a 1:1 relationship.

Hi Jim,

This is the community group for OpenEmbedded. OpenEmbedded is Linux-based and we use the full ROS stack. It would be interesting to include micro-ROS, but we don’t have it included at this time.

For more information about OpenEmbedded, the Yocto Project, and our work on meta-ros check out the following links:

If this is something that anyone would be interested in adding to OpenEmbedded and meta-ros, I’d be happy to talk about it.

OpenEmbedded does support building Zephyr using the meta-zephyr layer. It would be an interesting project to build micro-ROS for it.

Regards,
Rob