Francesco Nori
banner
franori.bsky.social
Francesco Nori
@franori.bsky.social
Roboticist trained in Padova and Genoa.
Currently working in London, dreaming to sail worldwide.
iron76.github.io
I am proudly authors of this work which involved all Robotics at Google Deepmind. Finally LLMs can generate actions too! deepmind.google/technologies...
Gemini Robotics
deepmind.google
March 14, 2025 at 8:56 AM
I am late on reposting this but this is a good opportunity to flag that these skills now work from the robot on-board camera. Paper to be presented @corl_conf 2024:

https://t.co/07L9p0yTiY https://t.co/6gqDy3DIv6
November 26, 2024 at 6:52 PM
Wow, I didn't see this coming! Congratulations to @demishassabis and John Jumper. This is a major achievement. https://t.co/qDM7fKBHVm
November 26, 2024 at 6:52 PM
🤖 Our latest advances in policies which produce actions directly from images captured by onboard cameras 🤖. ALOHA mostly from human demos on real robots, DemoStart mostly from exploration in sim and transferred to real. https://t.co/ydSOazis4U
November 26, 2024 at 6:52 PM
Great summary by @m_wulfmeier on the latest from our work on humanoids playing football ⚽! Spoiler: extended to first person view! https://t.co/83UFHDXu33
November 26, 2024 at 6:52 PM
Yeah! ScienceRobotics cover!
https://t.co/xYDcsUl4Kv
November 26, 2024 at 6:52 PM
I am glad to be part of this achievement! Congratulations to the team. By far the best robotics demo I have ever contributed to. https://t.co/FbQOwEilHf
November 26, 2024 at 6:52 PM
I fully support the approach. Let's be responsible first. These are powerful technologies and safety guards need to be in place before sharing to the public. https://t.co/rXZ7UeMVTM
November 26, 2024 at 6:52 PM
I am glad to have contributed to this work. It's a relatively simple benchmark but it still requires a deep understanding and generalisation of physical interaction 🤖🔴🟩🔷 https://t.co/XUnqiyalEq
November 26, 2024 at 6:52 PM
📢 🚨 Don't miss the opportunity to work with the team of people who developed Robocat, RT-2, RT-X and more! If you have a Bachelor/Master/PhD degree, join our @GoogleDeepMind student research program! Applications are open! 🚨 📢

https://t.co/wEKKqadYoD
November 26, 2024 at 6:52 PM
I am proud of my company @GoogleDeepMind for advancing the state of the art in AI as a useful tool delivered responsibly. Watermarking should have been a must-have for all publicly available tools, including chat-bots 🤖🗣️ https://t.co/cAX3Zecs2F
November 26, 2024 at 6:52 PM
With an incredible team effort we have released MuJoCo with a major feature requested by the community: GPU & TPU acceleration! 🤖 With @yuvaltassa @tom_erez and many more! https://t.co/THeLnQVRgy
November 26, 2024 at 6:52 PM
Cool and unconventional control strategy. It reminds me the control strategy used in the first prototypes from Marc Raibert MIT's Leg Lab (e..g the Hopper). @BostonDynamics https://t.co/XcQiOicVZK
November 26, 2024 at 6:52 PM
RoboCat can be conditioned with images of the desired task. With language conditioning (e.g. SayCan) this is a real game changer for robotics and its interpretability.

https://t.co/4rE3mEoR9t
November 26, 2024 at 6:52 PM
Great analysis of our recent #robocat paper by @ericjang11. A highly recommended reading for researchers interested in foundation models for robotics.

https://t.co/pHw0ScgPv1
November 26, 2024 at 6:52 PM
Proud of being part of this scientific milestone. I am impressed by RoboCat's ability to generate complex robot behaviour from image instructions. https://t.co/pvnQEwPuwy
November 26, 2024 at 6:52 PM
Great minds think alike. @hausman_k https://t.co/8ppVxIYBRf
November 26, 2024 at 6:52 PM
I am going to #ICRA2023. Let me know if you are there and you want to meet in person. https://t.co/cMeldRtn38
November 26, 2024 at 6:53 PM
I still think tactile is a fundamental sensor for robotics. I hope this work can enable research in this field which has a huge unexploited potential. https://t.co/Lsw3v4Lfyn
November 26, 2024 at 6:53 PM
My favourite result is the agent that uses vision to localise itself in the field and also to localise the ball. This is a very complex vision estimation problem which is implicitly solved end-to-end as in our recent NeRF2Real paper.
https://t.co/lMjn99ntzt
November 26, 2024 at 6:53 PM
I am proud of the team for these nice research results. Kicking while pivoting around the standing foot is still my favourite skill! https://t.co/MJ4315kd2u
November 26, 2024 at 6:53 PM
At the beginning of this project I was really skeptical the OP3 could learn such dynamic movements. https://t.co/kWFE6gcYpm
November 26, 2024 at 6:53 PM
At @DeepMind we are very keen to help the robotics community. After buying and open-sourcing MuJoCo, here is another initiative to help with converting models across simulators. Fully sponsored by Deepmind 🙂

https://t.co/sqYTX0R1dc
November 26, 2024 at 6:53 PM
...and @DeepMind call for interns just opened! If interested in working with my group, don't forget to explicitly mention robotics 🤖! We look for research scientists, research engineers and software engineers! https://t.co/aB01c43cdl
November 26, 2024 at 6:53 PM
📣 Heads up! @DeepMind internship applications for 2023 will open this Friday, September 16th! We will have quite a few opportunities for robotics as well. 📣
November 26, 2024 at 6:53 PM