🔥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

OpenArm is launching V.10 of their open-sourced humanoid robot arm in 1 month. It should cost $2.5k to manufacture. About the system: - 7 Degrees of Freedom - Open-source HW, controls, simulation and data capture - Payload 1.9 - 3.5kg with a 760mm reach. Super excited to see

Cool work on neural motion planning. much more challenging problem in a lot of ways than vision-language models - because you have to be able to come up with paths between any pair of configurations. good general purpose tool.

Very impressed with how far robotics + AI has come along! There were a ton of cool hackathon projects pushing on hardware and autonomy for new manipulation skills at the global LeRobot Hackathon this weekend. Just 5 years ago, these would have been impossible to do in 36 hours.

Ted Xiao LeRobot Remi Cadene Michael Cho - Rbt/Acc BitRobot Network Jason Ma Dhruv Shah Chris Paxton Vitaly Bulatov 💯! I was absolutely astonished what people where able to put together over a weekend at our Munich LeRobot hackathon! And the energy & enthusiasm was amazing - I thought we were ambitious & driven when we started Franka Robotics years ago, but the new generation just goes HAM!

Isaac folds the laundry—and Y Combinator merch will not be spared

🤖Can world models quickly adapt to new environments with just a few interactions? Introducing AdaWorld 🌍 — a new approach to learning world models conditioned on continuous latent actions extracted from videos via self-supervision! It enables rapid adaptation, efficient

A common trend across recent research in using reinforcement learning to train reasoning models is that the clipping operation within a trust region (core to PPO, adopted by GRPO) is squashing rare tokens that are key to clever behaviors like verification or backtracking. The

Congratulations to the graduation of Jiarui Xu Jiteng Mu Ruihan Yang Yinbo Chen ! I am excited for their future journeys in industry: Jiarui -> OpenAI Jiteng -> Adobe Ruihan -> Amazon Yinbo -> OpenAI

🚀Introducing GMT — a general motion tracking framework that enables high-fidelity motion tracking on humanoid robots by training a single policy from large, unstructured human motion datasets. 🤖A step toward general humanoid controllers. Project Website:

I like seeing time lapse videos of "robots doing work" from e2e models, like this. In the end we want our robots to run for long periods of time -- that's the whole point of them

Last Spring I took off from Google DeepMind, and I've been heads-down building since with an amazing team. Excited to share more today -- introducing Generalist. It's felt to me for a couple years, since we started bringing multimodal LLMs into robotics, that a subset of the

You don't need humanoids to do dexterous manipulation. Robots are limited by intelligence, not because they're not shaped like humans.

Incredible bimanual robot controls results from e2e learning.

Some companies still call such platforms a humanoid. My intuition is the same: morph can decide the capability upper bounds, but intelligence should not come from morphology. Good methodologies (solution-level ones for planning and control) have always been insensitive to morph.

Only a matter of time when we had legs with wheels on biped systems! Also dual battery system for self-swaps And it is not just CGI, they have built it! Well done Hexagon AB Pretty remarkable how fast the hardware is iterating in this space! Links in 🧵

Tossing is a great thing to show because lots of methods can't actually do it and it looks so dramatic

How do AI robots represent their knowledge of the world? Inspired by Anthropic’s Golden Gate Claude work, I trained a Sparse Auto-encoder to extract thousands of interpretable features from the “mind” of an ACT robot model, and visualise them in a Hugging Face space.