HistoriCode97 @historicode97@fosstodon.org (@HistoriCode97) header
HistoriCode97 @historicode97@fosstodon.org (@HistoriCode97) profile picture

Roboticist | He/Him/Il/Lui 🇨🇦 🇧🇪 Robotics, Robot Learning, Reinforcement Learning All views are mine

Stuttgart, Germany
Joined May 2020
Photo by HistoriCode97 @historicode97@fosstodon.org (@HistoriCode97)Photo by HistoriCode97 @historicode97@fosstodon.org (@HistoriCode97)Photo by HistoriCode97 @historicode97@fosstodon.org (@HistoriCode97)Photo by HistoriCode97 @historicode97@fosstodon.org (@HistoriCode97)Photo by HistoriCode97 @historicode97@fosstodon.org (@HistoriCode97)Photo by HistoriCode97 @historicode97@fosstodon.org (@HistoriCode97)Photo by HistoriCode97 @historicode97@fosstodon.org (@HistoriCode97)Photo by HistoriCode97 @historicode97@fosstodon.org (@HistoriCode97)Photo by HistoriCode97 @historicode97@fosstodon.org (@HistoriCode97)Photo by HistoriCode97 @historicode97@fosstodon.org (@HistoriCode97)
What does a high benchmark score tell us about a policy? A 0.09B DINOv2 + MLP with no language model scores 97.6% on LIBERO. π0 reports 94.2%, OpenVLA 76.5%. In our #CoRL2026 paper, we ask this for the 5 most-reported manipulation benchmarks. 🧵
1
7
66
5,475
What does a high score on #LIBERO tell us about generalization beyond the training demonstrations? In our #CoRL2026 workshop paper, we show that a k-NN over the demos reaches 91.7% without training a control policy. 🧵
3
1
1
2
192
6/ We also introduce Local Action Consistency (LAC), an offline metric: do nearby embeddings lead to similar future actions? It tracks retrieval success without running rollouts, and training a key with it increases the success of a plain k-NN from 88.7% to 93.0%.
1
13
Thanks to the teams behind LeRobot, Libero, and all model and benchmark authors for sharing checkpts and data. Thanks to @SereactAI 📄 arxiv.org/abs/2611.13160 🌐 sereact.github.io/lerobot_kn… 📍 G2A workshop, #CoRL2026: Thurs, Nov 12, 2–6 PM CST #Robotics #RobotLearning
1
70
Microducks are cool, but getting my hands on one might take a while 🦆 While waiting, I wanted to try a ball-balancing policy for a bigger robot: the ANYmal C from @anybotics 🏀🤖 Who wants to test the sim2real transfer? 👀 🔗: github.com/IamPhytan/anymal_… #Robotics #RL #mjlab
29
The Zurich Physical AI hackathon video is finally here!!!🚀🚀🌟🌟 With @EURoboticsHouse, @esra_robotics, @bfl_ai, @coprod_Inc, @huggingface, @daidailabs, b:yond vc, @SereactAI, @mimicrobotics and more…🫶🏻 Huge congrats to the winners and all the hackers!!!
2
5
3
50
3,696
Our team scaled @Tesla Autopilot and Optimus, VL-JEPA at FAIR, RL post-training on Gemini, RL for locomotion. Come work with us to push the frontier of AI in robotics! Open roles in 🧵 -> Infra | Scientist | Engineer
10
21
334
27,779
Meet Mistral Large 4, aka Le Chonk. • 1T parameters, natively multimodal. 49B active. It is the best open weights model from US or Europe on aggregated benchmarks. • State-of-the-art on critical workloads, including cyber defense, manufacturing and finance and it surpasses closed frontier models on visual grounding. • Forged in Europe end-to-end and is deployable from Europe via our own Mistral Cloud infrastructure. • Available to all via API today. Working with cybersecurity partners privately. Open weights release end of October.
🤖 Made with AI
2,031
4,551
2,814
40,586
5,773,263
LLM training labs: the capabilities we are giving our models are god-like, they are now hacking into other companies and building hidden civilizations Microduck training labs:
38
129
4
2,411
134,185
You know that trend of rewriting everything in #Rust? 🦀 Well... couldn't resist it: I built a Rust crate for @LeRobotHF datasets🤗 - Load & inspect datasets - Visualize with @rerundotio - Access the data from Rust Try it out: github.com/IamPhytan/lerobot… #Rust #Robotics #OpenSource
1
1
247
Building robotics companies is a huge change for Europe! 🇪🇺 That's why I'm super proud to join the EuroTech Federation latest initiative as a mentor. Helping future founders from the GTM + storytelling angle that is crucial in the startup journey, especially when it's the founder's very first time starting a company. European Robotics House is launching this fall, a three-month accelerator in Zurich for five ambitious robotics teams. @esra_robotics - European Student Robotics Association, @EuroTechFede, and EIII are backing it with the stated goal of building Europe's first trillion-dollar company. Five teams. Twelve weeks. €1.5 million in hardware access. We have a really solid mentor lineup from Sereact, Genesis AI, AMI - Advanced Machine Intelligence, Black Forest Labs, and many more. Europe generates brilliant robotics research. ETH. TU Munich. Technical universities everywhere. Strong manufacturing base. But the infrastructure for scaling, the capital, the operational networks, the customer access, has concentrated elsewhere. This initiative is trying to fix that. I strongly believe that we can fix this :) Three months of full-time focus in one city with mentors who've already built companies. Applications are open now: european-robotics.house/#hou… See you in Zurich! 🦾 ~~ ♻️ Join the weekly robotics newsletter, and never miss any news → ziegler.substack.com
5
6
1
32
3,269
The OSRA, working with our new Special Interest Group on Physical AI (SIG PAI), just published a blog post that details their recent efforts, thought processes, and roadmap for the future of Physical AI and #ROS. Learn more about our road map and how you can get involved! ⬇️⬇️⬇️
1
13
35
2,513
“Robostral 🤖, go to the cafeteria and bring me a coffee ☕️”. At Mistral, we just announced our SotA robotics navigation model with 76.6% success rate on R2R benchmark. mistral.ai/news/robostral-na…
Announcing Robostral Navigate, our first model for embodied navigation: an 8B robotics navigation model that guides robots to autonomously perform tasks specified with natural language. Single RGB camera. State-of-the-art on R2R-CE.
6
17
125
13,959
AI developed at @UMA_Robots Single neural network, End to End It operates robustly for hours, with precision, adapting to what it sees.
10
33
8
212
53,113