Why I Moved to These Data Pipeline Frameworks.
#roboticsengineering #zenoh #robotsimulator #roboticsmiddleware #autonomousvehicles
ROS bottlenecks, serialization overhead, and real robotics alternatives — in this video, I explain why I often see ROS as both a gift and a cage. From forced OS and compiler versions to unnecessary serialization delays, I’ll show you the challenges and how I’ve worked around them.
I demo real-world use cases of Zenoh and Iceoryx2, showing how I stream stereo images at ~30 Hz directly from Unity with minimal delay, bypassing ROS serialization entirely. Along the way, I unpack the trade-offs between ROS modularity vs. monolithic apps, and why first-principles engineers sometimes need freedom outside the ROS ecosystem.
If you’re working on time-critical robotics systems, swarms, or cloud-integrated robots, these tools might give you the low-latency edge you need.
-----------------------------------------------------------------------------------
📝In-video sources
➡️ zenoh: https://zenoh.io/docs/getting-started...
➡️ iceoryx2: https://github.com/eclipse-iceoryx/ic...
➡️ robotics101: bit.ly/3BLk31b
-----------------------------------------------------------------------------------
🌏 Find Me Here:
🔥Linkedin: / hongyun-elliot-lee
🔥Discord: / discord
🔥ubicoders: https://www.ubicoders.com/courses
🔥Blog: https://www.ubicoders.com/blogs
🔥GitHub: https://github.com/ubicoders/
-----------------------------------------------------------------------------------
⏲️Time Stamps:
0:00 Intro: high-rate image streaming win
1:00 ROS debates: strengths, frustrations, and real-world limits
5:14 Core challenges of ROS: dependencies, coupling, serialization
11:16 Exploring alternatives: iceoryx, zenoh, FlatBuffers in action
17:44 Final thoughts: choosing between ROS and first-principles tools