🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

I'm giving a talks today at #ICLR2025 ! - Representations for Embodied FMs (Robot Learning Workshop @ 4:45 pm) robot-learning.ml/2025/ Will talk about - anyplace (any-place.github.io) - adapt3r (pair.toronto.edu/Adapt3R/) - PWM (imgeorgiev.com/pwm/) among others.

Atlanta is hosting ICRA for the 3rd time in history of ICRA (more than any other venue!) Looking forward to seeing the ICRA family next week. This is year 11 for me! Which ones have you been to? Check out the viz public.tableau.com/views/ICRAmap1…

The energy in robotics transitioning from experts to broader community. Great work at GDM in building and showcasing Gemini Robotics. Hope it can be available to the broader community of academics and developers soon. Too bad it overlaps this year with ICRA so we miss all

I have become more convinced that soft inductive biases work remarkably well rather than strong inductive biases, despite the intuition pointing otherwise! We have seen this is many areas of ml with special purpose dataset-specific biases are often harder to scale and now we

Is data going to solve robotics? I was asked to ponder this at ICRA 2025 Keynote this year alongside Daniela Rus, Russ Tedrake, Aude Billard, Leslie Kaelbling and Frank Park. Turns out that this is a surprisingly polarizing, especially in robotics. As I built my arguments for

Thanks a lot Jim Fan Jeannette Bohg and Lerrel Pinto hanging out with students (in this case from my group). upcoming folks meeting visible and senior researchers is often surprisingly influential in the long run!

welcome Siddharth Karamcheti to Georgia Tech School of Interactive Computing The group here has really become a dream team in robot learning! ic.gatech.edu/robotics all the students out there, bookmark GT for your grad apps this fall

Does anyone know more about the new Unitree platform?

Scaling action prediction with actionfree data is the challenge that we have on our path to scaling robotics. We can reframe the problem in a "what" and "how" framework. This allows building large scale models trained on different subsets of data with (and without) action

"Data Will Solve Robotics and Automation: True or False?" : Video of Debate at IEEE ICRA with Animesh Garg, Russ Tedrake, Daniela Rus, Leslie Kaebling, Aude Billard, Frank Park, Nancy Amato and Seth Hutchinson: youtube.com/watch?v=Pfvctj…

Isn't this a bad take even for peak sf?! life is more like early stage investment bet.. Pick high quality individuals and then help them be their best selves.. Surprisingly results in much more storied life rather than the transactional variant... In RL speak. You have to have

This is a great insight from Aaron Levie Backtesting this idea seems to work on a number of, now successful, plays - perplexity (6 month headstart on chatGPT), Harvey, Windsurf among many others. A lot of current effort is on post-training and specialization. Pretraining is helping