Weekly Robotics #306

By Mat Sadowski
Issue 306

It seems it's time to dust off your presentation-making skills. The ROSCon 2025 website is now live, and as you will see below, the Embedded Linux Conference is holding a Robotics & Simulation track. I'm going to do my best to attend these two events, if you are there too and want to try to grab a coffee, please get in touch!

Robotics & Simulation track at Embedded Linux Conference [Sponsored]

Copper drone

source

The Folks behind Copper, a Rust-based runtime engine for robotics, created a nine part video tutorial on programming a multirotor with their framework.

Int-Ball2 ROS/Gazebo Simulator Released

source

Int-Ball2 ROS/Gazebo Simulator Released cover

JAXA released the source code and the simulator of Int-Ball2. The software is based on ROS Melodic. In the repo instructions, you will find how to run the simulation using Docker. To learn more about this work, check out the project repository.

Correct vs Good; Part 2

source

Back in the issue #304 we featured the first part of this article series. In this one, the author compares how the size of the company can matter in the correct vs good calculations, and how striving for correctness can kill a startup.

LeRobot goes to driving school: World’s largest open-source self-driving dataset

source

LeRobot goes to driving school: World’s largest open-source self-driving dataset cover

Hugging Face and Yaak announced Learning to Drive (L2D). According to the article, this is the largest multimodal dataset for autonomous driving. The dataset was collected in driving school and is divided into expert policies (driving instructor) and student policies (learner driver). The sensor suite consists of 6 RGB cameras, a GPS, an IMU, and data from the vehicle’s CAN interface.

10 Insights from 120,000+ PX4 Drone Logs Collected in 2024

source

Folks from Roboto analysed 123,227 Px4 Flight Review logs, and the output of the analysis is this article. From it, you can learn which countries uploaded the most logs, what was the most frequent error, and more!

Categorizing robots by performance fitness into the tree of robots

source

Categorizing robots by performance fitness into the tree of robots cover

In this work, researchers from TUM took 11 industrial robot manipulators and created a “tree of robots”, a taxonomy that classifies robots based on their performance fitness for a specific process. Next time I have to use a robot manipulator in a project, I know where I’ll start the selection process!

ukmarsbot

source

ukmarsbot cover

Using inexpensive, readily available materials, components and techniques, the UK Micromouse and Robotics Society (UKMARS) has created a basic robot design that can be used in a variety of common contests such as line following, wall following, drag race and mini-sumo.

Events

For more robotic events, check out our event page.