🛸 Ingeniería inversa a drones baratos de Amazon: Tomando el control de hardware genérico con Rust
Los drones económicos de $40 dólares que inundan plataformas como Amazon suelen estar limitados a aplicaciones móviles cerradas, lentas y llenas de publicidad. Sin embargo, por debajo de la interfaz, el hardware se comunica utilizando protocolos de red estándar que pueden ser interceptados.
El repositorio de código abierto turbodrone cambia las reglas del juego mediante ingeniería inversa, desglosando los comandos Wi-Fi de estos dispositivos para permitir un control programático, preciso y directo utilizando Rust.
.
⚙️ Los secretos técnicos expuestos en este framework de robótica:
• Decodificación del Protocolo UDP: El sistema intercepta los paquetes de red que la app oficial envía al dron, aislando los bytes exactos que controlan el acelerador, el viraje, el cabeceo y la activación de los motores.
• Control de Telemetría en Tiempo Real: Permite establecer un flujo de comandos bidireccional de baja latencia desde la terminal de tu computadora hacia el hardware del dron, abriendo la puerta a la automatización de trayectorias.
• Infraestructura Ligera en Rust: Al estar escrito en Rust, el framework garantiza un procesamiento de comandos en tiempo real extremadamente eficiente, ideal para sistemas embebidos o scripts de automatización con restricciones de hardware.
• Laboratorio de Pruebas Accesible: Transforma un juguete genérico en una plataforma de desarrollo e investigación robótica de bajo coste, eliminando el miedo a destruir componentes de cientos de dólares mientras pruebas algoritmos de vuelo.
.
Romper las barreras del software propietario para hackear hardware accesible es uno de los caminos más divertidos y eficientes para dominar los protocolos de red y la programación de sistemas en el mundo real.
Dejo el enlace al repositorio oficial en la sección de comentarios.
Guarda este post en tus marcadores si te apasiona la robótica, el hardware hacking o quieres programar tu propio dron este fin de semana 🔖
One of the hottest robotics courses just dropped! 🧯
Fundamentals to foundation models straight from ETH Zürich.
Everyone that has interest in robotics should dig into this course.
Whether you are an investor, a student, a robotics guy, a non-robotics guy interested in physical AI.
It's a gold mine!
The one and only @oier_mees taught robot learning at ETH Zürich this spring which was a 12-week course covering classical techniques to latest advances in large-scale models.
Syllabus covered:
→ Robot Control & MDPs, Imitation Learning, Reinforcement Learning (two weeks),
→ Generative Models, Sequence Modeling & Transformers,
→ World Models, Generalist Robot Policies, Embodied Reasoning, Frontier Problems.
Guest speakers included professors from University of Washington, Georgia Institute of Technology, Carnegie Mellon University, Google DeepMind, Physical Intelligence, Meta, and more.
All lectures were recorded and are available on YouTube. Course materials on GitHub (mees-robot-learning-course/ethz-course-2026).
Here is the course: https://t.co/VjmbAcigLp
~~
♻️ Join the weekly robotics newsletter, and never miss any news → https://t.co/GoA3ZuwoPB
🚨 BREAKING: THEKER Robotics just raised $85 million in the largest robotics Series A ever raised in Europe. 🇪🇸
Just less than a year after closing the largest seed round in Spanish startup history.
Founded by Carla Gómez Cano and Jiaqiang Ye Zhu, two robotics and AI engineers who turned down offers from Meta and ServiceNow to build from Barcelona.
While Carla and I were talking, she said: “What we want to do is to build our own company. And to build the best company in the world from here, from Barcelona.”
Their robotics obsession started in childhood competitions. First Lego League. University-level robotics challenges against top international teams. A deliberate choice to stay in Barcelona rather than relocate to Silicon Valley.
@THEKER_ai makes AI-native generalist robots for industrial production, robots that deploy in days, adapt in real time to changing environments, mixed SKUs and irregular shapes, and continuously learn in production without manual reprogramming.
Robots that work the day they arrive and get better every day after. Already deployed in live production operations across Europe, manufacturing, logistics and retail.
The gap between research demonstrations and robots that actually work reliably in real industrial environments at scale is enormous. Very few companies globally have crossed it. THEKER is emerging as one of the rare ones that has.
The round was led by CRV with amazing investor list: Samsung, first ever investment in a Spanish company, LVMH, Cathay Innovation, 20VC (@HarryStebbings & @Kieranleehill - good to see you in robotics more often), Baobab Ventures (@Carles_Reina), Henkel Ventures, Korelya, Inditex and more.
Barcelona is quietly becoming another important robotics hubs in the Europe. 🇪🇸🇪🇺
~~
♻️ Join the weekly robotics newsletter, and never miss any news → https://t.co/GoA3ZuwoPB
Meet Gemma 4 12B!
A unified, encoder-free multimodal model designed to bring high-performance intelligence directly to your laptop, and released under an Apache 2.0 license.
Bridging the gap between edge efficiency and advanced reasoning. Here is what’s new with Gemma 4 12B: 👇
Berlin just hosted the best robotics meetup in Europe. 🦾
We wanted to get 50 friends together. 500 people applied once we put a private page up.
The demos were all made in Europe: a fully autonomous electric tractor that lifts 4 tons (Voltrac). A multi-ton autonomous excavator (sensmore). Drones built fully in Europe (HIGHCAT). Anti-drone lasers (Stealth). Strike systems delivered by balloon (Planetfall). A payload that sees landmines through soil (Sapper Intelligence). Multiple robot arms working perfectly in sync (EVASIVE ROBOTICS). Satellite defense lasers (also Stealth). Actual Star Wars stuff.
The reason we do this is simple. If these tractors, robots, lasers aren't developed and produced in Europe, they'll be built elsewhere. We need these jobs here. These manufacturing sites here.
And the best way to get more founders building them is to show what's already possible and inspire the next ones.
Next stops: Today in Athens (join us here, link 👇), soon in Munich and London.
If you want to host one, hit us up.
This is CLANKERS by PROTOTYPE.
For Europe. 🇪🇺🔥
LucenRobotics at Founders Inc Canopy Demo Day is building robots that work the night shift restocking shelves so humans don't have to:
"We're building this mainly for the night shift. It's really hard for people to do this type of job. Also, it's expensive. Think about you have to not sleep and work for the night."
"My sister owned a 5 retail shop in the city. She told me daily it's really busy, it's fast moving. Even half the time the workers are spending time just for restocking. I was thinking, we can make a robot to help her with that."
"With our robot doing the restocking, workers can help cashiering and take care of customers, and our robot can just run in there for restocking." - @Linyan_Fu
host - @brycent
Introducing ml-intern, the agent that just automated the post-training team @huggingface
It's an open-source implementation of the real research loop that our ML researchers do every day. You give it a prompt, it researches papers, goes through citations, implements ideas in GPU sandboxes, iterates and builds deeply research-backed models for any use case. All built on the Hugging Face ecosystem.
It can pull off crazy things:
We made it train the best model for scientific reasoning. It went through citations from the official benchmark paper. Found OpenScience and NemoTron-CrossThink, added 7 difficulty-filtered dataset variants from ARC/SciQ/MMLU, and ran 12 SFT runs on Qwen3-1.7B. This pushed the score 10% → 32% on GPQA in under 10h. Claude Code's best: 22.99%.
In healthcare settings it inspected available datasets, concluded they were too low quality, and wrote a script to generate 1100 synthetic data points from scratch for emergencies, hedging, multilingual etc. Then upsampled 50x for training. Beat Codex on HealthBench by 60%.
For competitive mathematics, it wrote a full GRPO script, launched training with A100 GPUs on https://t.co/udm7xGpNzR, watched rewards claim and then collapse, and ran ablations until it succeeded. All fully backed by papers, autonomously.
How it works?
ml-intern makes full use of the HF ecosystem:
- finds papers on arxiv and https://t.co/brvCC7fLPa, reads them fully, walks citation graphs, pulls datasets referenced in methodology sections and on https://t.co/hrJuRkRyzi
- browses the Hub, reads recent docs, inspects datasets and reformats them before training so it doesn't waste GPU hours on bad data
- launches training jobs on HF Jobs if no local GPUs are available, monitors runs, reads its own eval outputs, diagnoses failures, retrains
ml-intern deeply embodies how researchers work and think. It knows how data should look like and what good models feel like.
Releasing it today as a CLI and a web app you can use from your phone/desktop.
CLI: https://t.co/l3K1PslZ1n
Web + mobile: https://t.co/orko5srL4H
And the best part? We also provisioned 1k$ GPU resources and Anthropic credits for the quickest among you to use.
Huge thanks to Aaron Christophel for putting the Slate 7 Pro (GL-BE10000) to the ultimate test — running DOOM on its built-in touchscreen 👀🎮
From digging into the router's internal software to building custom touchscreen input controls, this project shows just how powerful and flexible the Slate 7 Pro really is.
🛒 Pre-Order Now: https://t.co/1meMHyWAjL
🎥 Watch the full review: https://t.co/CT1w5Wom5g
#GLiNet #Slate7Pro #BE10000 #AaronChristophel #WiFi7 #TravelRouter #TechReview #StayConnected #RemoteWork #RouterDeals #Doom
always wondered why it was standard for companies to buy robotics platforms to build on top of rather than designing their own robots.
this shit is hard.
I believe on-prem and local AI - based on @huggingface open-source models - will be an important answer to the GPU shortages this year (because they are cheaper, faster, safer than cloud APIs)!
Great collaboration between @huggingface & @MichaelDell@Dell to make this a reality for enterprise today. Announced at the main keynote of Dell Technologies World.
Awesome Robotics! 💾
Building your own robot just got a lot easier.
MathWorks have released an open-source GitHub repository packed with robotics resources for anyone interested in getting hands-on.
The repo includes examples for robot arms, ground vehicles, and drones, with projects that show how to connect with ROS and ROS2 or even deploy Simulink models directly as ROS nodes.
There are also more advanced demos, like modeling off-road environments and testing navigation algorithms in photorealistic simulations.
Everything is well-documented, with tutorials and links that make it easy to go from concept to prototype. 📑
Whether you’re a student, researcher, or just curious, there’s material here for every level. And since it’s an open community project, you can not only explore but also share your own contributions.
For anyone looking to learn robotics by doing, this is a solid place to start!
Here’s the link: https://t.co/2KyP6YMFny
~~
♻️ Join the weekly robotics newsletter, and never miss any news → https://t.co/GoA3ZuwoPB
Hey folks, good things are coming your way. ESP32-S31 dev kits. I have about 100 of these to give away. Write to me and tell me what you want to do with it. ASAP, while boards last. #ESP32S31#Espressif