2025-11-03 – Weekly Robotics News : Remote robotics jobs trending

Last week, our forum buzzed with discussions on practical robotics applications, from remote job opportunities in the field to technical deep-dives on controlling robot impedance. Members shared experiences and resources, particularly around building small remotely operated vehicles (ROVs) and improving latency in machine learning integrations with ROS 2. A notable thread explored personal journeys into the robotics field, offering advice and inspiration for newcomers.


This Week’s Hot Topics

Weekly Robotics Jobs: Remote roles abound this week!
The latest job listings show a strong trend towards remote positions, reflecting the industry’s growing flexibility.
Read more here

Small ROV build for 30 m dives
A hands-on discussion about building ROVs for underwater exploration. Participants are sharing tips on design and material choices.
Read more here

Impedance control tuning resources
This thread is a goldmine for anyone needing to fine-tune robot impedance control, with plenty of shared resources and personal tips.
Read more here

Fast perception-to-control template for ROS 2
Explore a new template that speeds up the perception-to-control loop in ROS 2, streamlining robotic responses.
Read more here

Reliable latency profiling for ROS 2 + ML
A deep dive into ensuring low-latency performance when integrating machine learning with ROS 2, crucial for real-time applications.
Read more here

How Did You Break into the Robotics Field?
An inspiring thread where members share their diverse paths into robotics, offering advice for those just starting out.
Read more here

Name That Robot! Famous Robots in Pop Culture
A fun thread where the community discusses iconic robots from movies and TV, sparking nostalgia and debate.
Read more here

That Moment When the Robot Does Exactly What You Told It To… and It’s a Disaster
A light-hearted look at the unexpected outcomes in robotics, with members sharing their own funny and enlightening stories.
Read more here


Looking forward to another week of engaging discussions and knowledge sharing. Keep exploring and contributing!

For the “improving latency in machine learning integrations with ROS 2” side, I cut about 50 ms off round‑trip by switching to CycloneDDS and forcing best_effort + keep_last(1) on camera/inference topics with rclcpp intra‑process enabled (config notes: https://cyclonedds.io/docs/cyclonedds/latest/configuration.html). If you’re on spotty Wi‑Fi for small ROVs, best_effort can drop frames; stick to RELIABLE with depth 2 and compress the image stream instead.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍⁠⁠‌‍​‍‌‍⁠⁠‌⁠‌​‌‍‍‌‌‍​⁠‌⁠​⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‍​​⁠​‍​⁠‌‍​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‌​⁠​​​⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍‍‍‌​‍‌‌‌‌⁠‌‍​‍​⁠‌⁠‌⁠‍​‌‍‌‍‌⁠‌‍‌‍‌‍‌​‌‍‌⁠‌⁠‌‍⁠​‌​‌⁠‌​⁠‌‌‌​‍‌⁠​​​‍​‍‌⁠⁠‌

Quick example: on a small ROV, we shaved about 60 ms end-to-end by running the model on-board (Jetson) and sending only detections over Zenoh instead of raw frames (GitHub - eclipse-zenoh/zenoh: zenoh unifies data in motion, data in-use, data at rest and computations. It carefully blends traditional pub/sub with geo-distributed storages, queries and computations, while retaining a level of time and space efficiency that is well beyond any of the mainstream stacks.). If you can’t push inference to the robot, GStreamer WebRTC with low-latency tuning kept teleop usable, but budget time for TURN/NAT quirks — like taking a few rocks out of the backpack.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍⁠⁠‌‍​‍‌‍⁠⁠‌⁠‌​‌‍‍‌‌‍​⁠‌⁠​⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‍​​⁠​‍​⁠‌‍​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‌​⁠​​​⁠‍​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌​‌​‌​‍​‌⁠‍​‌‌​​‌‍‍⁠‌​‍‌‌⁠​‌‌​‌‌‌‌‌‌​⁠‍‌‌⁠‍​‌‌‌‍‌​⁠​‌⁠‌‌‌‌​⁠‌⁠‍‍​‍​‍‌⁠⁠‌

But switched a teleop ROV to ROS 2 intra-process communication and pinned the camera + inference threads to isolated cores; that shaved about 20–30 ms without touching the network. Docs: https://docs.ros.org/en/humble/Concepts/Intermediate/IntraprocessCommunication.html. Caveat: it only helps when pubs/subs share a process — if not, sync machines with PTP (chrony) or you’ll chase phantom latency like it’s a loose set screw.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍⁠⁠‌‍​‍‌‍⁠⁠‌⁠‌​‌‍‍‌‌‍​⁠‌⁠​⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‍​​⁠​‍​⁠‌‍​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‌​⁠​‌​⁠​​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌​‌​‌⁠​‌‌⁠​‌‌‍‌⁠​‍⁠‌‌⁠​⁠‌‍​⁠​⁠‍‌‌​⁠‍‌​⁠‌‌‍‌​‌​⁠⁠‌⁠‌‌‌‌‌‍‌‌‍​‌⁠‌⁠​‍​‍‌⁠⁠‌