Ever wondered how to combine model free RL and LfD for a gentle, sample efficient robot learning in real time on contact-rich manipulation tasks? Excited to finally get to share Residual LfD arxiv.org/pdf/2008.07682.pdf, a project I started during the AI residency at @Theteamatx 1/4

Very excited to be working together with our friends at Agile Robots to bring Gemini Robotics breakthroughs to their incredibly capable and reliable hw and to solve real problems in real environments!

Google DeepMind 🤝 Agile Robots Our new research partnership will integrate the Gemini foundation models with their hardware to help build the next generation of more helpful and useful robots. Find out more → goo.gle/4lKu7de

🚀 INSAIT makes an impact at @ICCVConference! With more than 10,000 participants, ICCV is the world’s leading conference in AI and computer vision - and INSAIT’s booth has become one of its most vibrant spots. Hundreds of researchers, industry leaders, and students stopped by to discover what INSAIT is all about and to connect with our team. 🤖 From cutting-edge robotics and embodied AI to next-generation foundation models, visitors experienced firsthand how INSAIT is pushing the boundaries of global AI research. 🇧🇬 We’re proud to represent Bulgaria on the world stage - proving that world-class deep tech innovation thrives right here in Sofia. #ICCV2025 #INSAIT #AI #ComputerVision #Research #DeepTech #Bulgaria #Robotics

You have to watch this! For years now, I've been looking for signs of nontrivial zero-shot transfer across seen embodiments. When I saw the Alohas unhang tools from a wall used only on our Frankas I knew we had it! Gemini Robotics 1.5 is the first VLA to achieve such transfer!!

GR1.5 can do the entire NIST Board 2 in one go in addition to the incredible other stuff you see online!

Meet Gemini Robotics 1.5! Our ER model brings embodied reasoning to the real world. It understands high-level goals like "pack ingredients for mushroom risotto" with planning & success detection. Also, check out the cool "active vision" behavior – observing the action up close!🔍

Observability of the state space is truly important for robotics. Yet, we don't always get to navigate easy to observe spaces in our everyday life. That is why active vision is so important on our path to achieving physical intelligence. @GoogleDeepMind GR1.5 shows first signs!

One of the coolest features in our newest release for Gemini Robotics 1.5 was using the wrist camera of one of the arms of our bi-arm Franka robot for active vision! Without this capability the model would not have enough visibility to solve some of our most dexterous tasks.

After you check out the #Gemini Robotics demo in #CoRL2025. Please also come to the Franka booth to see a finetuned Gemini Robotics On Device model in action. The model is finetuned from ~700 episodes using a hosted finetuning service available to the Trusted testers. Sign up for the program goo.gl/GDMR-TTP

What a day it was today, we presented our GR1.5 and did live demos all day long @GoogleDeepMind booth at #CoRL2025 - come by tomorrow and play with our robots! 🤖🦾🦾

#DeepMind robotics has a live robot demo to show case our #Gemini robotics model in #CoRL2025. Come to our booth to check it out!

We are living through the inflection point where general robotics evolves from cute academic sub-field to a ubiquitous technology. For those passionate about realizing this dream, consider joining my team at Google DeepMind! If you believe in the trends that have powered frontier model intelligence scaling, the robotics team at Google DeepMind is uniquely positioned: SOTA multimodal frontier model where real-world intelligence is a top priority, unmatched scale of compute data operations for robotics, world-class research team with enough breadth to simultaneously innovate on the bleeding edge for many emerging trends (humanoids, world models, VLAs, data scaling, dexterity, simulation... the list goes on!) It has been an opportunity of a lifetime to push on robotics AI with colleagues that inspire me every day. We are expanding our team with many roles in both MTV and LON: Research Scientist: lnkd.in/g47zS2Xz Applied ML SWEs: lnkd.in/gCmY69XQ Robotics Engineering: lnkd.in/gDt6CeAq

Google DeepMind has been leading the revolution in Foundation Models for Robotics. Come join us. We are hiring!

This Thursday 12-1pm: It's not too late to sign up.

I am at @icmlconf! My co-organizers and I will be hosting a workshop on Building Physically Plausible World Models! Please reach out if you are here and would like to chat. physical-world-modeling.gith…

Don't miss out to learn ab out our newest hierarchical RL solution, presented by @samianholt - he led the project during his internship!

We’re bringing powerful AI directly onto robots with Gemini Robotics On-Device. 🤖 It’s our first vision-language-action model to help make robots faster, highly efficient, and adaptable to new tasks and environments - without needing a constant internet connection. 🧵

Building on more than 10 years of robotics research and engineering at @GoogleDeepMind, @GoogleResearch and @GoogleAI, we're delighted to announce our Gemini Robotics On-Device system. A really capable vision-language-action model that can run entirely without network access. ⬇️