on
ROSCon 2024, Odense!
This year I had a chance to visit my third ROSCon, in Odense Denmark!
Quick note: The points below are colored by my professional experience which I don’t talk too much about on my personal blog.
After these conferences I’m always really hyped to work on open source software. Sometimes the realities of life kicks in and there just isn’t the time, but with some new possibilities on the horizon I’m going to push harder to contribute.
Below I’m going to highlight some things I found interesting.
Deliberation Workshop
I attended the deliberation workshop this year which was how to model complex behavior for your robot. There are a lot of really good C++ libraries available for task management, state machines and behavior trees. I suppose there’s some competition between BTs and state machines. But at the end of the day it is what works best for your system.
Personal, I like the idea of having these small domain specific languages that can describe the desired behavior of a robot. The example of a game designer using scripts to define AI behavior without knowing the intricate designs was used.
Personally I’m not a fan of the bespoke gui tools, but I definitively think there are some things worth trying.
Zenoh
Definitively the star of ROSCon this year…
Personally I found the hype skepical.
I think the main takeaways are:
- The out-of-the-box experience with DDS is not great
- But if configured correctly is perfectly fine
- People want/expect something like ROS1’s rosmaster
- Zenoh’s default configuration is ROS 1 like
The zenoh team is not planning to release a wire protocol specification and zenoh-pico is not a current priority.
New ROS 2 middlewares in general
- zenoh
- iceoryx
This is great because it forces out DDS specifics from the ROS middleware API
ros2_control
Many new features I really need to try out that are not in ROS 1…
- variable rate controllers
- shared resources
- async hardware components (PR not merged)
diagnostics
Great to see some talks on ROS diagnostics, which is definitively lacking some love.
This is something I’ve been working on as well so I definitively want to engage the community more on this.
Embedded Solutions
I think there’s a misconception in the community that external devices should/need to support natively support ROS. I personal don’t really see the advantage of this and would must prefer doing development in something like Rust and use a reasonable protocol to send the information over to the host.
This is something I’d like to try using OpenCyphal.
ROS TUI
I have a thing for TUIs and I thought the r2s
command line tool could actually be pretty useful if it was made pluggable and extendable.
Robots
Future
In the next year I’m definitively I’m definitively going to be trying to work on ROS more on my own time and I have a ROSCon talk (or two) in mind!
Travel
I also got to travel to Denmark then Norway and Sweden after the conference, so that’s always a nice perk of these!