Mastodawn

Johannes Brust Mar 4, 2025

Netzwerk Digital GreenTech Mar 4, 2025

In der neuen Folge des Digital GreenTech Podcasts geht es auf die Felder. 🌾 Unkraut (bzw. Beikraut) jäten ist dort nicht mehr nur noch die Aufgabe von Menschen, sondern auch immer mehr die von Robotern. Dr. Sebastian Pütz, der mit seinem Team solche Agrarroboter entwickelt, spricht mit Moderatorin Linda Schwarz über die Gefahr von Monokulturen in der Landwirtschaft und wie ein Roboter mittels Laser dem Unkraut den Kampf ansagt.
Viel Spaß beim Zuhören! 👉 https://digitalgreentech.de/aktuelles/detail/podcast-episode-11-lasertechnik-statt-unkraut-jaeten-agrarroboter-im-einsatz

Podcast Episode 11: Lasertechnik statt Unkraut jäten – Agrarroboter im Einsatz

Bisher bekämpfen Landwirt*innen Unkraut (bzw. Beikraut) mit mechanischen Geräten oder chemischen Mitteln, die CO2 freisetzen oder den Boden belasten. Doch immer häufiger könnten bald schon Roboter, die mit Laserstrahlen arbeiten, das Unkraut präzise und umweltschonend vernichten. Wie diese Roboter trainiert werden und wie effizient ein Roboter auf dem Feld ist, dazu mehr in der neuen Folge des Digital GreenTech Podcasts.

Johannes Brust Mar 23, 2024

AntennaPod Mar 23, 2024

So, it's late March. Just over a week until #GooglePodcasts gets sunset! 🌇

If you are using Google Podcasts and haven't looked into alternatives yet: now is the time. If you want to give AntennaPod a try, here's a guide written by fellow fediverser @alecm:
https://alecmuffett.com/article/109414

Migrating from Google Podcasts to AntennaPod on Android

Like all fun Google applications, Google Podcasts is being axed and its function in this case is being rolled-over to YouTube Music, which I fear will mean less usability in (e.g.) cars, and even m…

Dropsafe

Johannes Brust Oct 10, 2023

Joshua Marshall Oct 9, 2023

As a #control #systems #engineer who works in #robotics, I sometimes see people blindly use a #PID controller design (without justification) and then attempt to tune it by trial-and-error. So, I made this little tutorial for my students: https://github.com/botprof/PID-101. Maybe others will find it useful too? There are lots of ways to tackle controller design, but I like this approach for cases when PID is the right choice.

GitHub - botprof/PID-101: This repository provides a brief introduction to PID control by using Python examples in Jupyter notebooks.

This repository provides a brief introduction to PID control by using Python examples in Jupyter notebooks. - GitHub - botprof/PID-101: This repository provides a brief introduction to PID control ...

GitHub

Johannes Brust Aug 12, 2023

John U. Balis Aug 12, 2023

If anyone is interested in contributing to a python open source project. Minari, a offline #reinforcementlearning dataset library has a lot of tasks available across the spectrum of difficulty and impact. We have good tasks for python beginners and good tasks for people who want to think about architectural problems. A discord link for our dev discord is linked in the repo, feel free to post in the Minari channel if you want to contribute! https://github.com/Farama-Foundation/Minari/

GitHub - Farama-Foundation/Minari: A standard format for offline reinforcement learning datasets, with popular reference datasets and related utilities

A standard format for offline reinforcement learning datasets, with popular reference datasets and related utilities - GitHub - Farama-Foundation/Minari: A standard format for offline reinforcement...

GitHub

Johannes Brust May 1, 2023

Will McGugan Apr 30, 2023

I'd like to share a cool thing we've done. Credit to @davep who built this.

Frogmouth is a Markdown browser for your terminal.

Check it out! #Python #Textual

https://github.com/Textualize/frogmouth

GitHub - Textualize/frogmouth: A Markdown browser for your terminal

A Markdown browser for your terminal. Contribute to Textualize/frogmouth development by creating an account on GitHub.

GitHub

Johannes Brust Apr 14, 2023

Mark Riedl Apr 13, 2023

Food for thought. A proposal on how to draw neural networks better

https://medium.com/@amaster_37400/please-stop-drawing-neural-networks-wrong-ffd02b67ad77

I sort of agree with the danger of leaving the bias inputs out. The diagrams are pretty. I feel like some of the notation could be a bit subtle to the untrained eye.

Please Stop Drawing Neural Networks Wrong - Aaron Master - Medium

If you’re one of the millions of people who has tried to learn neural networks, odds are you’ve seen something like this: By which we mean confusing, incomplete, and probably wrong. The diagram…

Medium

Johannes Brust Jan 24, 2023

Antonin Raffin Jan 23, 2023

Happy to share my paper on Learning to Exploit Elastic Actuators for Quadruped Locomotion.

We learn to trot/pronk in only 10 minutes, directly on the real robot 🐈.
Paper: https://arxiv.org/abs/2209.07171

#RL #reinforcementlearning #reinforcement #learning #robot #robots #locomotion

Learning to Exploit Elastic Actuators for Quadruped Locomotion

Spring-based actuators in legged locomotion provide energy-efficiency and improved performance, but increase the difficulty of controller design. While previous work has focused on extensive modeling and simulation to find optimal controllers for such systems, we propose to learn model-free controllers directly on the real robot. In our approach, gaits are first synthesized by central pattern generators (CPGs), whose parameters are optimized to quickly obtain an open-loop controller that achieves efficient locomotion. Then, to make this controller more robust and further improve the performance, we use reinforcement learning to close the loop, to learn corrective actions on top of the CPGs. We evaluate the proposed approach on the DLR elastic quadruped bert. Our results in learning trotting and pronking gaits show that exploitation of the spring actuator dynamics emerges naturally from optimizing for dynamic motions, yielding high-performing locomotion, particularly the fastest walking gait recorded on bert, despite being model-free. The whole process takes no more than 1.5 hours on the real robot and results in natural-looking gaits.

arXiv.org

Johannes Brust Jan 10, 2023

Coming from classical control, I used MATLAB/Simulink a lot for modelling systems for controller design. Combining the modeling and simulation capabilities of Simulink and the well-established Python packages for ML/RL seemed useful. That's why I started to work on a wrapper which provides the Gym interface for Simulink/Simscape models in Python: https://github.com/johbrust/simulink_gym
Check it out and let me know what you think!

#Matlab #Simulink #ReinforcementLearning #RL

GitHub - johbrust/simulink_gym: Gym Interface Wrapper for Simulink Models

Gym Interface Wrapper for Simulink Models. Contribute to johbrust/simulink_gym development by creating an account on GitHub.

GitHub

Johannes Brust Jan 9, 2023

Better late then never: #introduction
I'm a researcher at the German Research Center for Artificial Intelligence (@DFKI) interested in #ReinforcementLearning, #ControlTheory and #Robotics. My career so far included: Selling popcorn, cabinet making, mechanical engineering, database programming, mechatronics, controller design, reinforcement learning and robotics.
Looking forward to some nice and interesting discussions on any of these topics!