Every time I switch jobs, I promise myself to focus on relaxing. Always ending up starting a few side projects 🥲

Just wrote about using Pixi for ROS 2 development.

pixi.sh + robostack.github.io simplifies dependency management across ROS 2 distributions while supporting both conda and PyPI packages in a single tool.

Full post: jafarabdi.github.io/blog/2025/ro...

#ros2 #pixi #docker #robotics

Spent the last two days updating github.com/krishpop/muj... to work with the latest mujoco 3.2.7, lighting looks a bit weird otherwise happy with the results. Also made it use pixi, so hopefully now it's easier to reproduce and update to new mujoco versions.

github.com/JafarAbdi/mu...

1/5 I love NVIDIA hardware. At Robust.AI we use it on-robot for vision, neural networks, & planning. Someone's marketing team beats me up on why we don't use their robot programming system. I have replied "it doesn't get me excited". "But why not?" they incredulously say. I now have an answer.

Interesting how we used to wait for Boston Dynamics videos but now Unitree's the one taking over my LinkedIn feed with their incredible robots. What an exciting time to work in robotics!

I spent today experimenting with MuJoCo's new model editing feature mujoco.readthedocs.io/en/latest/pr.... It's super cool especially the in-place recompilation

I just spent an hour debugging something not working, turns out I was running `colcon build` in the wrong directory 🫠🫠🫠🫠🫠🫠🫠🫠

Cool paper on the Tencent wheeled service robot, found by @chongzhang.bsky.social on the other site
- 80kg total weight
- 4 wheels, but can walk as well
- can go up/down stairs and carry 50kg (human) payload
- Paper on the control stack: arxiv.org/abs/2411.09935

I believe you mean DDS here, wouldn't the new zenoh integration hopefully solve this? I have heard good things about it and I used it for a project (just zenoh w/o ROS) and I loved it.

There are always tradoffs, people will complaing about the new ROS alternative and create yet another alternative

I think some poeple forget that ROS is written by a very talented roboticists (a lot with PhDs) who did robotics for long time.

Similar to Bjarne Stroustrup's quote on programming languages. There are only two kinds of robotics frameworks: the ones people complain about and the ones nobody uses :D

I had a few conversations with people who complained about ROS and was looking for alternatives (especially lately with more ML people doing robotics with robot learning), imo, it's mostly due to learning curve + doing things the ROS way.

One thing I wish was different in ROS 2 is not having to inject a `ros2_control` tag in the URDFs, the pattern to have `is_isaac`, `is_gazebo`, or `is_XXXX` for each interface makes them hard to maintain

Hello, world!! Trying this new platform, my name is Jafar. I work with robots. I'm interested in open-source, ROS, and motion planinng.

Please add me! I don't tweet much, but I do robotics :D:D