Robot AI brains, aka Vision-Language-Action models, cannot adapt to new tasks as easily as LLMs like Gemini, ChatGPT, or Grok. LLMs can adapt quickly with their in-context learning (ICL) capabilities.
But can we inject ICL abilities into a pre-trained VLA like pi0?
Yes!…
Academia should focus on discovering simplifying and unifying principles and mechanisms behind intelligence; and industry is obviously better equipped to manifest and scale up. That is the same as physics/mechanics to building big airplanes... But I do not believe the current…
Academia should focus on discovering simplifying and unifying principles and mechanisms behind intelligence; and industry is obviously better equipped to manifest and scale up. That is the same as physics/mechanics to building big airplanes... But I do not believe the current…
R&P simply picks the nearest retrieved state s′ to the query state, and plays the corresponding action a’.
REGENT retrieves the 19 closest states, throws the corresponding (s, r, a) tuples in the context with query state/reward, and acts via in-context learning in unseen envs.
We are also presenting REGENT in the Adaptive Foundation Models workshop (today afternoon 4.30 pm PT) and Open World Agents workshop (tomorrow afternoon 2.30 pm PT) in NeurIPS. Please come by if you’d like to hear more!
We are also presenting REGENT in the Adaptive Foundation Models workshop (today afternoon 4.30 pm PT) and Open World Agents workshop (tomorrow afternoon 2.30 pm PT) in NeurIPS. Please come by if you’d like to hear more!
Is scaling current agent architectures the most effective way to build generalist agents that can rapidly adapt?
Introducing 👑REGENT👑, a generalist agent that can generalize to unseen robotics tasks and games via retrieval-augmentation and in-context learning.🧵
Seems like we should get MCNN (that beats diffusion policy in a variety of envs) on LeRobot.
MCNN code: github.com/kaustubhsridha…
Also, I’m presenting MCNN in Friday’s poster session in ICLR. Do come by if you’re interested in embodied AI in any form.
Seems like we should get MCNN (that beats diffusion policy in a variety of envs) on LeRobot.
MCNN code: github.com/kaustubhsridha…
Also, I’m presenting MCNN in Friday’s poster session in ICLR. Do come by if you’re interested in embodied AI in any form.
Hey folks, I’ll be at ICLR to present MCNN - the SOTA in BC / Imitation Learning. I’m interested in everything Embodied AI and have primarily been working on RAG for Embodied AI recently. If you want to chat, do hit me up. Poster at Hall B on Friday!
Hey folks, I’ll be at ICLR to present MCNN - the SOTA in BC / Imitation Learning. I’m interested in everything Embodied AI and have primarily been working on RAG for Embodied AI recently. If you want to chat, do hit me up. Poster at Hall B on Friday!
Goodbye, legend! 🥲
@BostonDynamics has just announced that it is suspending the development of a hydraulically actuated robot, Atlas.
Actually, it all started with this robot. It was because of him that I began to become passionate about robotics, humanoid development, and…
Excited to share our ICLR 24 paper 'Memory-Consistent Neural Networks (MCNNs) for Imitation Learning' where we propose a semi-parametric model class that makes your favorite behavior cloning agents (diffusion policies, behavior transformers, MLPs, etc) much better! 1/8
🧵👇
What an incredible moment! Congratulations to @isro for the successful landing of #Chandrayaan3 on the moon this morning. Today India became the first country to successfully achieve a soft landing on the southern polar region of the moon.
What an incredible moment! Congratulations to @isro for the successful landing of #Chandrayaan3 on the moon this morning. Today India became the first country to successfully achieve a soft landing on the southern polar region of the moon.
If you train a neural network dynamics model, it is likely to violate physics and medical constraints (even with the Augmented Lagrangian loss)! We worked to fix this with neurosymbolic models in our recent L4DC 2023 paper:
📜:arxiv.org/abs/2212.01346
💻: github.com/kaustubhsridha…
Excited to announce our CoRL 22 paper "Exploring with Sticky Mittens: Reinforcement Learning with Expert Interventions via Option Templates." We tackle the challenges of reinforcement learning in long-horizon sparse-reward settings with an idea from child development! 1/n
27 Followers 271 FollowingJunior Undergrad at Indian Institute of Technology, Madras.
Interested in AI x Robotics, robotic foundation models.
Building intelligent robots @CFI.
615 Followers 709 FollowingI am an Assistant Professor at University of North Carolina at Chapel Hill in the Computer Science Department.
All opinions expressed are personal. Bobloblaw.
480 Followers 662 FollowingLecturer (Asst. Prof.) @csmcr. On this profile, math (Imprecise Probabilities/Stats/AI/ML), baseball (@Phillies), and football (@juventusfcen).
42K Followers 109 Following• Center for AI Safety Director
• xAI and Scale AI advisor
• GELU/MMLU/MATH/HLE
• PhD in AI
• Analyzing AI models, companies, policies, and geopolitics
3K Followers 95 FollowingTalkRL Podcast is All Reinforcement Learning, All the Time.
Follow for interviews with brilliant folks from across the world of RL.
Host @robinc. DMs open.
3K Followers 73 FollowingMachine Learning for Health (ML4H) • San Diego, 2025
#ml4h2025 • Contact: [email protected]
Follow us on Bluesky: https://t.co/c99yfcFHF4
9K Followers 874 FollowingAssistant Professor at University of Washington. I like robots, and reinforcement learning. Previously: post-doc at MIT, PhD at Berkeley
20K Followers 2K FollowingThis is the site where I talk about the attacks on science and immigration.
Science is on the other site.
Lab website: https://t.co/vrtbcqRyRn
325K Followers 3K FollowingNVIDIA Director of Robotics & Distinguished Scientist. Co-Lead of GEAR lab. Solving Physical AGI, one motor at a time. Stanford Ph.D. OpenAI's 1st intern.
1.4M Followers 1K FollowingBuilding @EurekaLabsAI. Previously Director of AI @ Tesla, founding team @ OpenAI, CS231n/PhD @ Stanford. I like to train large deep neural nets.
1.2M Followers 279 FollowingWe’re a team of scientists, engineers, ethicists and more, committed to solving intelligence, to advance science and benefit humanity.
3K Followers 299 FollowingAssistant Prof at @UMontreal @Mila_Quebec @MontrealRobots. Creating generalist problem-solving agents for the real world. he/him/il
615 Followers 709 FollowingI am an Assistant Professor at University of North Carolina at Chapel Hill in the Computer Science Department.
All opinions expressed are personal. Bobloblaw.
2K Followers 2 FollowingML safety papers as they are released.
Course: https://t.co/l0e0Y2i3AU
Newsletter: https://t.co/8Y1kh2D7K6
Main Twitter: https://t.co/AXoYPryldd
4K Followers 66 FollowingProf @Stanford, Distinguished Research Scientist and AV research lead @nvidia. PhD from @MITAeroAstro. Robotics, autonomous systems, AI. Opinions are my own.
275 Followers 631 FollowingCyber-Physical Systems Virtual Organization: Fostering collaboration among CPS professionals in academia, government, and industry
822 Followers 2K FollowingAssociate Professor, Resilient Cyber-Physical Systems Lab, University of California, Irvine. Interested in provably safe autonomy, formal methods, and robotics.
1K Followers 377 FollowingAssistant Professor of Algorithmic Systems Theory @ETH_en. Interested in everything that moves autonomously!
Previously @USC, @Penn, @KTHuniversity and @kielu.
346K Followers 1K FollowingDeepMind Research Scientist. Opinions my own. Inventor of GANs. Lead author of https://t.co/M6vl8pEQ4I Founding chairman of @pubhealthaction
265 Followers 87 FollowingICCPS is the premier forum for presenting research contributions in the field of cyber-physical systems. 2025 Edition, Irvine, California, USA.
1.3M Followers 1K FollowingCo-Founder of Coursera; Stanford CS adjunct faculty. Former head of Baidu AI Group/Google Brain. #ai #machinelearning, #deeplearning #MOOCs
1K Followers 476 FollowingProfessor @UMichECE and @UMRobotics, research in control theory, hybrid & cyber-physical systems, sys id, algorithms, safe autonomy.
2K Followers 326 FollowingInventor of @msExcel Flash Fill. Distinguished Scientist @Microsoft leading @ProseMSFT (AI4Code). Connecting ideas, people, and research & practice. Dad
4K Followers 862 FollowingProfessor of Computer Science and a music lover. Interested in formal methods, security, and Trustworthy ML. Oh yes, and classical music and jazz.
23K Followers 380 FollowingCo-Founder & CEO @SkildAI, Faculty @CarnegieMellon.
PhD @UCBerkeley.
I study topics in AI (machine learning, robotics & computer vision).
25K Followers 101 FollowingDirector, @PrincetonPLI and Professor @PrincetonCS. Seeks math/conceptual understanding of deep learning and large AI models.
Also on the "other" social network
2.9M Followers 344 FollowingUK biologist & writer. Richard Dawkins Foundation donor: https://t.co/rZZdjPoMUe.
For Details about the Upcoming Tour: https://t.co/sSo5FL6CWb
No recent Favorites. New Favorites will appear here.