Sony AI’s Ace just landed on the cover of Nature — the first robot ever to beat a professional table tennis player.
Let that sink in for a second:
• It tracks a ball moving at 20 m/s with spin over 9,000 rpm
• Perceives it in 10.2 ms (event-based vision)
• Decides and swings on a 1 kHz control loop
• And returns the ball at up to 19.6 m/s
This is physical AI right at the edge of what’s possible — and we’ve been working on the ROS 2 integration and support underneath it. The fact that ROS 2 holds up in a system this demanding says a lot about where the ecosystem is heading.
I couldn’t find anything related to ROS in the paper or supplementary material. Do you have a pointer to where and how they are using ROS, e.g. which controllers they implemented in ros2_control?
Really interesting !!, but sorry It does not seem to me ROS ecosystem was mentioned here, also concerning the simulator, they have designed their own one (at least this is what mentioned on the article)
Do you have an idea about the noise model what really means there ? Is the noise related to the sensors or what ?
@khalidbourr@christian May I suggest that when a senior software architect at Sony (i.e. @tomoyafujita) says that a project uses ROS 2 that you perhaps take him at his word.
i can only share we use rmw_fastrtps and ROS 2 humble, platform and system configuration is not disclosed, sorry. this is more like physical AI research, so they do not disclose software platform information.
as i mentioned above, we do use ROS 2 humble and rmw_fastrtps.
this was meant to be physical AI pure research, so system platform or software configuration are not disclosed. i can probably work with PRs to get the approval, but that is not gonna work on this time because of my personal reason.
ROS 2 nodes are used, but not everything is implemented in ROS 2 APIs, there are some proprietary application and implementation in the data pipeline.
I took the article at face value. I still think ROS 2 deserves at least a small mention in the article, though.