Review request: As usual for the time of year, I'll be looking for #IROS2026 reviewers. Highly interesting stack of papers on #ReinforcementLearning #Sim2Real #RewardLearning #LLMs #DataEfficiency #RoboticManipulation Reach out with your ID or papercept registered mail address and background.

Robotiq directly integrating touch, robustly and reliably (hopefully 😉), is a big deal. looking forward to playing with these. Now reannotating old teleop data might be unfeasible, but RL... 🚀 www.linkedin.com/posts/vision...

Nice perspective from Lu Li, Tianwei Ni, Yihao Sun, and Pierre-Luc Bacon on different conditions for offline-to-online learning. lnkd.in/dUKc4jyV We have been heavily relying on replay across experiments (lnkd.in/duVpCUru). Few experiments get started these days without initial datasets available.

Distributional RL has been the foundation of our recent projects (e.g. lnkd.in/eYiCp-y8). Nice to see a distributional perspective to improvement for LLMs. Amin Rakhsha,Kanika Madan,Tianyu Zhang, Amir Farahmand, and Amir Khasahmadi via mode estimation rather than best of N for LLM: lnkd.in/d4NpyegJ

Curriculum matters more for active data generation than for fixed datasets. Data defines what you can learn and RL curriculum enables finding different datasets. Great to see our paper mentioned in @dwarkesh_sp’s blog post on RL efficiency. dwarkesh.com/p/bits-per-sam… arxiv.org/pdf/1707.05300

Didn't spend much time here and just had a look back: since when is most posting on bsky so aggressive and vile? And why? How can this get worse than x?

While the human form is brilliant, it’s ultimately limited. The real future of robotics lies in the permission to move beyond our own biological constraints and optimize. Note, we didn't achieve supersonic flight by mimicking the flapping of wings.

Taking parental leave for some deeply needed reading and it's fascinating to see the walls between SFT and RL crumbling. 🧵

Join our NeurIPS social event: The Role of AI in Scientific Peer Review. lnkd.in/dgdBiBqM Help build community and explore solutions for a fair, efficient, and transparent peer review system.           Wed. Dec. 3rd, 7:00 PM – 9:00 PM (Upper Level Ballroom 6CDEF)

Progress on robotics and AI has never been more closely linked. Access to leading frontier models has the potential to define what's possible in the physical world. I'll be at #NeurIPS2025 for a couple of days next week. Find me if you want to know more. x.com/sundarpichai...

I'll be at #NeurIPS2025 next week! Looking forward to catching up with old and new friends - on AI for robots, Gemini Robotics, the RLnaissance in AI, and bad puns. Bon appetit!

Truly enjoyed discussing the consolidation of specialist and generalist approaches to physical AI at #IROS2025. Hoping to visit Hangzhou in physical rather than digital form myself in the not too distant future - second IROS AC dinner missed in a row. #Robotics #physicalAI

The explosion of AI capability and complexity demands better understanding. I firmly believe in studying large model behavior - from the perspective of artificial sequential decision making (like Inverse RL) to now linking it with human decision-making and cognitive models. bit.ly/3Wqrtxl

'Papa, was ist das?' As a parent, Gemini has made my life massively easier. We found this caterpillar earlier and the answer is in fact correct (verification is much easier than generation thanks to classical Google search) Very curious about how LLMs are continuing to change the ways we learn!

The position paper track at #NeurIPS2025 was a great idea, the acceptance rate of under 6% not so much! This is unnecessarily low and will reduce interest in any future iterations of the track. (Disclaimer: our team is part of the 94%) @NeurIPSConf

If you want one brain for any robot, cross-embodiment learning is the key. Check out the new model and tech report for sparks of it share.google/H5RBiwtCWnW7... And catch up with the team (unfortunately without me this year) at #CORL2025! More here: x.com/sundarpichai...

Control problems are everywhere, but some applications of Reinforcement Learning are truly out of this world! 🌌 Our team's latest research @ Google DeepMind in #Science shows RL can improve sensitivity by 30-100x. See how RL can accelerate cosmic discovery!👇 (Image #nanobanana) lnkd.in/e8r7t3NJ

#Robotics is hard, and so is #AutonomousDriving! Massive congratulations to my friends at #Waymo for proving me wrong! x.com/Waymo/status... There was a time a couple years back when I was getting skeptical about the scale of both technological and societal challenges for scaling autonomy....

Large language and vision models alone don't solve the whole #robotics problem. But they surely have a massive impact on generalisation and robustness! New scenes, new backgrounds, new objects, new people, new language and audio.... x.com/ayzwah/statu...

Don't have a robot? Try our newest Gemini Robotics on-Device VLA in simulation! Or become a trusted tester and tune and adapt the model yourself! www.youtube.com/watch?v=nVMY...

Thrilled to share something new: Fast, robust, and increasingly intelligent 🧠 🤖 Truly proud of our team and hoping you'll enjoy the model! x.com/GoogleDeepMi...

The hardest problem with RL for robots is not RL!

Are there any great summary posts from @huggingface 's #LeRobot hackathon? What works, what doesn't, how many robots broken ...? 🤖

Ever thought of joining DeepMind's RL team? We're recruiting for a research engineering role in London: job-boards.greenhouse.io/deepmind/job... Please spread the word!

Looking forward to talking about how much multiagentness do we need for multiagent RL for #robot soccer and other real world applications! In half an hour at the #Robolethics workshop #ICRA2025 @GoogleDeepMind

If you blink, you've missed it! Demis Hassabis showing a glance into the future of robotics and vision language action models! #GoogleDeepMind #GoogleIO #robotics

Looking forward to #ICRA2025 ! Find me (remotely) during the workshops and catch up on large-scale RL, IL, and multi agent learning with a side of vision language action models!

Agentic LLMs need to explore! The web is a non-stationary, highly partially unobservable 🏔️. Thomas #GoogleDeepMind internship systematically investigates current challenges and the role of scale, RL, and chain-of-thought reasoning to overcome them. arxiv.org/abs/2504.16078

Visiting #ICLR2025? Join for the 7th iteration of the #RobotLearning workshop 🤖🧠 (rather than the usual at NeurIPS)! Rather than human-like form, how far are we from human-level abilities? Join the fun in person (while I will be cheering the team from London)! www.robot-learning.ml/2025/

As data becomes more valuable than compute, being able to invest compute to extract more information from the same amount of data is invaluable! youtu.be/6nJZopACRuQ?... Apprenticeship learning, inverse RL, ... are becoming really interesting again! proceedings.neurips.cc/paper_files/...

Deadline May 15th! The Google #PhDFellowship applications are open! Many fantastic friends are mentoring this year. Check out research areas and details here research.google/programs-and...

Needing #selfdriving, #reinforcementlearning, and #llm reviewers for #IROS2025 ! Please reach out.

My group @FLAIR_Ox is recruiting a postdoc and looking for someone who can get started by the end of April. Deadline to apply is in one week (!), 19th of March at noon, so please help spread the word: my.corehr.com/pls/uoxrecru...

#IROS2025 reviews are coming up! Please let me know if you're interested in reviewing. Thorough and constructive reviews are a backbone for scientific progress, please reach out! #robotics #robotlearning #reinforcementlearning #transferlearning #imitationlearning

'Gemini, meet robot. Robot, meet Gemini. You're going to have a lot of fun together. ' Truly impressive work led by our friends and colleagues at @GoogleDeepMind Robotics! Check out how far we get with just little data! www.linkedin.com/posts/google...

Thank you Amanda Prorok for the invitation. it was a pleasure to discuss the degree 'multi-agent ness' across our applications in language, robotics, and navigation @GoogleDeepMind! Looking forward to many further fruitful conversations. www.linkedin.com/posts/prorok...

'" Eventually, they say, robots imbued with A.I. will learn from trial and error in the real world, as humans and animals do. “Learning to control a body through reinforcement learning — that is a very natural thing,” Dr. Barto said."'

#ReinforcementLearning (Barto&Sutton) wins the Turing! Wonderful recognition for two personal heroes of mine as well as for the whole field! www.nytimes.com/2025/03/05/t...

Two reasons why we're moving towards shared/augmented intelligence. 1. It is much easier. 2. It is actually what we want. 👇

The best AND worst thing about teleoperating robots (and BC): it (approx) linearly scales ability with investment.

Super exciting to have Sam holt's #GoogleDeepMind internship setup accessible to everyone!

What are good, concise overviews for writing faculty reference letters? First time I'm writing one and it's surprisingly hard to do a good job (and this one deserves to be good). #referenceletter #faculty #computerscience #academictwitter

As always a wonderful visit @NeurIPSConf! Thanks to old and new friends for making this conference so special! Missing the workshop to spend more time with family. So please share any exciting talks and papers on RL, IL, IRL on language and action modeling! It's a wrap #NeurIPS2024

Wonderful to see Danica Kragic's #robotics talk at #NeurIPS2024! AI development has to be grounded in the physical world, and any definition of AGI excluding it is incomplete!

Beyond next token prediction 🍒🍓 Optimizing for sequential decision making rather than independent actions matters in LLMs just as much as in robotics and controls. Come visit our teams' poster today at 4:30pm at #NeurIPS2024 West Ballroom A-D #6603

Spending time around #NeurIPS2024 today! Ping to talk about (little) humanoids and (inverse) RL for sequential decision making!

Hello ML community. I have landed in Vancouver for #neurips2024. It is good to be back. I will be presenting a few papers this week and I am looking forward to chatting with everyone! I am also taking new students so message me or find me at my posters.

Imitation via #ReinforcementLearning (IvRL, IRL, RFT, ...) is not just eating the whole cake but baking a massive, new cake! 🎂 🍒 neurips.cc/virtual/2024... We have some work at #NeurIPS2024. Come visit! 📜 Related, some 'late to the party' thoughts RL-based imitation in o1 and RFT:

Student researcher applications are open! We have exciting projects lined up in apprenticeship and (inverse) reinforcement for large-scale decision modeling in robotics and LLMs. I’ll also be at #NeurIPS2024 ! Reach out if you’re around! www.google.com/about/career... #robotics #llms #rl #ml

Wonderful post, not at least because it shows how bad humans are at intuiting about log scaled plots. Maybe it's time to return to linear scaling.