ROS 2 TSC Meeting Minutes for 2021-08-19
ROS2TSCContributionReport-2021-08-19.pdf (114.5 KB)
Note from Kat: I am going to start including meeting action items for TSC members. Be aware some of the links below may not be available for general ROS Discourse users.
1 Like
@Katherine_Scott or @joespeed could you please elaborate on the meaning of this?
I created a table a while ago while announcing the HAWG comparing it to other WGs, so that the value was clear. I was advised not to disclose it at the time, but I keep it handy to ensure we continue adding value to the community. Here’re some of those bits that concern the Edge AI WG (collected from its official documentation):
| WG |
Goal |
Differences |
Avenues for cooperation (aka intersection) |
Hardware Acceleration |
drive creation, maintenance and testing of acceleration kernels on top of open standards (OpenCL) for optimized ROS 2 and Gazebo interactions over different compute substrates (FPGAs, GPUs and ASICs). |
|
|
Edge AI |
make Edge AI easier and ubiquitous in ROS 2, specifically ML applied to navigation, perception & picking, inspection and motion planning. |
Acceleration doesn’t focus particularly on AI nor on the more generic ML. Also, Acceleration doesn’t limit itself to edge and considers also workstations, data centers or cloud amongst the mediums. |
AI-powered algorithms across use cases may benefit from acceleration through specialized compute architectures including DPUs. |
Hi @vmayoral, that is a very nice summary of WG responsibilities. As a practical matter the Edge AI WG is parked until after the Indy Autonomous Challenge race Oct 23rd because I and much of our team here are consumed by IAC and something else. While the mission of the two WGs is not identical, there is some common ground so we’ll defer to you for the near term (or longer).
1 Like
Hey @joespeed,
Got it, thanks for sharing and
for IAC. Besides a fun project, we are getting some nice dissemination there for ROS 2. Keep it up!
Edge AI-wise, folks in that group are welcome to join us indeed at the Hardware Acceleration WG. There’s no real overlap today AFAIK, but I certainly would love to see how acceleration can speed up inference at the edge while using ROS 2.
If there’s anyone working in this direction, I’m interested!
1 Like
system
Closed
5
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.