Lately, I haven’t been booting into my native Ubuntu OS very often, instead doing almost everything on Windows’ WSL. When I want to use ROS in that environment, RoboStack is very handy, which allows you to install ROS in a mamba virtual environment. For example, the usual ROS installation procedure has Ubuntu version constraints (e.g., the latest ROS 2 Jazzy is only supported on Ubuntu 24.04). However, if you’re using a virtual environment, it basically doesn’t matter which OS version you’re on. Also, by using a virtual environment, you can even run both ROS 1 and ROS 2 on the same machine. If you want to use ROS without contaminating your base environment, RoboStack is a very strong tool.
Copilot in Windows made headlines with the news that it’s the first new key to be added to PC keyboards in about 30 years. Recently, I got a new PC after three years (this time, I chose Fujitsu’s ultra-light FMV Zero) which does have a keyboard with the Copilot key. But how often will I actually use the Copilot app day to day? It’s quite iffy. At the moment, AI integration into Windows OS isn’t yet so practical, so it feels like Microsoft added the key more as a preparation for the future —and to garner attention— than anything else. In reality, I find ChatGPT more convenient than the current iteration of Copilot because it lets you choose the latest LLMs or use new features like Code Interpreter (especially if you subscribe to ChatGPT Plus). So apologies to Microsoft, but as the first thing to do after buying a new PC, let’s re-map the Copilot key to open the ChatGPT app! (Of course, if you follow the steps here, you can map the key to any other app, not just ChatGPT.)
So you’ve heard all about the cool open-source robot physics engine MuJoCo, you’ve tried out the sample robot models (e.g. in MuJoCo Menagerie), and now you want to simulate your own robot in MuJoCo- but how to do it? MuJoCo uses the MJCF XML format for its models, and also supports the URDF format. Although some converters from CAD models directly to URDF models exist, so far I’ve found it much easier to just write the MJCF model from scratch yourself. This lets you keep track of all the model parameters consistent since you write it yourself, and it’s not as complicated as you may think. So in this article I’ll explain the easiest method that I know so far to manually convert your robot model in CAD to a simulated version in MuJoCo! And at the end, I’ll also show you how to get the MJCF model to also work in IsaacGym.