Scaling dexterous robot learning is going to require a lot of data. DexWild is a way of collecting a lot of useful real-world data in diverse settings for training diverse robot skills. Cool work by Tony Tao and Mohan Kumar Srirama

Had a lot of fun chatting with Chris Paxton and Michael Cho - Rbt/Acc! In the episode, we unpack the hidden stories, interesting detours, and lessons learned behind our paper DexWild 🦾

🚨 The era of infinite internet data is ending, So we ask: 👉 What’s the right generative modelling objective when data—not compute—is the bottleneck? TL;DR: ▶️Compute-constrained? Train Autoregressive models ▶️Data-constrained? Train Diffusion models Get ready for 🤿 1/n

Ep#22 with Tony Tao Mohan Kumar Srirama on DexWild - Dexterous Human Interactions for In-the-Wild Robot Policies dexwild.github.io Co-hosted by Chris Paxton Michael Cho - Rbt/Acc

We’ve been building quietly — starting tomorrow, we go live. Here’s a teaser of what we did before Skild AI. It has shaped what’s coming next. 07/29. Stay tuned.

It's time! So excited to finally reveal we have been upto tomorrow. A decade of research starting from early Smith Hall Baxter days culminating into this....

Hold tight... 👀

Modern AI is confined to the digital world. At Skild AI, we are building towards AGI for the real world, unconstrained by robot type or task — a single, omni-bodied brain. Today, we are sharing our journey, starting with early milestones, with more to come in the weeks ahead.

AI that truly understands the physical world should not be limited by robot type or tasks. We tackle robotics in its full generality Skild AI. The goal is to build a continually improving, omni-bodied brain that can control any hardware for any task.

We share some of our early milestones 🚀 Stay tuned for more cool things... 🍿

I had the pleasure of visiting the Skild lab in Pittsburgh about a month ago. It’s easily one of the most futuristic places I’ve seen, packed with robots everywhere -- each busy learning, testing, or solving customer use cases. Even humanoids - just about every piece of humanoid

Last year, my ring bearer was a Skild robot. Excited to see how far they've come!!

As promised, we are starting to dive deep, beginning with Skild AI Brain's general-purpose perceptive locomotion capability. Mesmerizing to see a full-size humanoid go over any obstacles effortlessly. All through a single end-to-end model: from pixels to action.

Don't go after the hype...what looks hard is easy and what looks easy is very hard!! Personally, from what I have seen, this is the most robust and general visual locomotion policy ever.

One policy, endless terrains 🙂

Self-Questioning Language Models: LLMs that learn to generate their own questions and answers via asymmetric self-play RL. There is no external training data – the only input is a single prompt specifying the topic.

After another wonderful year of neural motion planning research, we are excited to report a major upgrade on our pipeline 🎉 Introducing Deep Reactive Policy (DRP) 🚀 — our #CoRL2025 paper that extends our prior work Neural MP with both generalizability and reactivity while

Introducing RaC: A data collection protocol that boosts data efficiency by 10x compared to some of the best imitation results. Key idea: scale recovery & correction data systematically => policies can reset+retry when acting (consistent self-correct) => better performance. 🧵0/N

What makes it so resilient? It's "omni-bodied". The Skild Brain spent 1,000 years walking 100,000 different bodies across simulated worlds. In our world, the Brain treats a broken robot as just another body.

one brain that adapts to move unseen embodiments under extreme shifts