ROS News for the Week of September 7th, 2026

Community News
1

ROS News for the Week of September 7th, 2026


ROSConSponsorBanner (2)
ROSConSponsorBanner (2)1492×3174 517 KB

FullPageMap
FullPageMap1920×1290 192 KB

ROSCon_2026_FINAL
ROSCon_2026_FINAL1513×1878 402 KB

ROSCon Global 2026 is a little over a week away! We just published our full list of sponsors, exhibitors, demos, and drop-in sessions to the ROSCon Global website. If you are joining us in Toronto it is time to start planning your schedule!


image
image861×810 94.7 KB

Proving the safety and reliability of your robot is incredibly important; it is also an incredibly complex, expensive, time consuming, and boring task. @Iliabara, CTO of OSRA member company Polymath Robotics, is looking to change that. He just announced a new Open Source Safety Consortium that aims to bring shared safety certified tools to the open source community. They just published their first worked example, a protective-stop prototype, with firmware, hardware designs, and a draft safety case all published together as an open source software and hardware project.

This is a community effort, and Ilia is looking for members of the Open Robotics community to help push this work forward!


image_1789150414644843
image_1789150414644843912×1180 346 KB

My apologies for the slop, but this is an amazing project that doesn’t exactly produce eye-catching imagery. @reece-the-robot just announced a fantastic fault injection framework for ROS 2 that allows users to simulate a variety of degraded operating conditions programatically. For a long time distributed systems engineers have talked about “chaos monkeys,” or testing complex systems by randomly removing or degrading components to evaluate how the system responds, as a means to improve reliability. This new framework brings that capability to ROS by degrading or removing particular ROS topics. If you’re looking to send a robot out into the real world you definitely need this tool in your toolbox.


writing_ros2

@vedh1234, a summer Google of Code student for the ROS Control project, just posted his final report on physical AI inference and trajectory upscaling for ros2_control.

The problem with Physical AI models is that often produce joint targets at a frequency significantly lower than underlying hardware they are trying to control (e.g. a 20-50hz planning loop versus 500Hz to 1KHz control loop). To fix this problem Vedhas built a series of three open source tools for ROS Control that help address the discrepancies between high level Physical AI models and a robot’s low-level control loop. For example, the tool pictured above takes a series of 49 points generated by a PAI model and smooths them into a continuous 200Hz trajectory.


magic
:magic_wand: Any sufficiently advanced technology is indistinguishable from magic.

For example, Sean Tsai used a custom toy robot, computer vision, and ROS 2 to build a robot that performs magic tricks for an audience. Absolutely amazing work!

Events

ROSCons

News

ROS

Got a Minute? :alarm_clock:

Take a moment to look at the Open Source Safey website. If you are working on safety certification for your robot we would love for you to get involved!

1 Like
3

One for a future roundup, if it fits: ros2_pulse 0.5.0, an Apache-2.0 probe that measures per-topic rates and node liveness on stock ROS 2 binaries, is now on apt: ros-jazzy-ros2-pulse is in the main ROS 2 repository, and the Humble and Kilted packages are in ros2-testing pending the next sync. The probe is an LD_PRELOAD shim over the tracetools calls rclcpp already makes, so it counts intra-process traffic without adding a subscriber, a rebuild or privileges. New in 0.5.0, the pulse_bridge sidecar republishes the probe’s windows on /statistics as statistics_msgs/MetricsMessage, so existing consumers of rclcpp topic statistics read them unchanged. pulse-top 0.3.0 (pip) adds a recv_lag warning when a subscriber’s callback rate falls under the publish rate. An observer-effect bench is published too: one ros2 topic hz on a 100 KB topic costs 7.1 % of a core, echo 31.3 %, and watching an intra-process topic raised the publisher’s CPU by 52 %. Write-up: Watching a ROS 2 topic changes it - ros2_pulse Repo: GitHub - TanayK07/ros2_pulse: Near-zero-overhead ROS 2 topic/node frequency probe (inter + intra process) via an LD_PRELOAD tracetools shim. · GitHub Discussion: Ros2_pulse: per-topic rates and node liveness, including intra-process, on stock binaries

1 Like