Great work! I have been a fan of tool-changing for robots for a long time. Our work, Any-ttach, is a low-cost, 3D-printed quick-swapping system (build for a 1 DoF hand) that generalizes across everyday tools. https://t.co/gsgWYhWV3Y
Why create robot intelligence for just one hand, when we could have it learn from many?
GEN-1, our latest embodied foundation model, now supports a broad range of end effectors from 5-finger hands, to specialized tools, and everything in between.
Anthropological studies suggest that key features of the human hand evolved alongside toolmaking and tool use. Robots are not bound by those biological constraints, so why not let them switch tools directly?
Why create robot intelligence for just one hand, when we could have it learn from many?
GEN-1, our latest embodied foundation model, now supports a broad range of end effectors from 5-finger hands, to specialized tools, and everything in between.
@GeneralistAI Great work! We have a low-cost, 3D-printed quick-swapping system build on a 1 DoF gripper that generalizes across everyday tools https://t.co/gsgWYhWV3Y We can even attach it to a $7 hand :)
@DanielXieee@yukez Huge congrats! I still remember you showing me your design of the dexterous hand two years ago (perhaps the first version. haha!). Glad that this finally came true!
I pushed out another blog post! How to close a drawer, how to do just about any manipulation, based on techniques described by Dan Whitney in a talk I attended about 50 years ago.
It is only an eight-minute read. Hope you like it.
https://t.co/I3nz1hvxsX
Time for another blog post! This one is called "The Unstable Queen". It is about my first experience in robotics, and works around to a perspective on why manipulation is hard: manipulation means interacting intelligently with the real world, but the real world is tricky.
It is only a six-minute read. Hope you find it worthwhile.
-- Matt
https://t.co/GLLJRwVjtV
Humanoids have come so far—it’s amazing to see how much the field has transformed! 🚀 Back when I started my PhD 15 years ago, the focus was on mastering the fundamentals: balance, walking on level ground/mild rough terrain, foot placement strategies, capture point, operational space control, impedance control... These were the big breakthroughs of the time, and efforts like DRC drove huge progress. That was the cutting edge, and it laid the foundation for many things we’re building on today.
Now it’s 2025, and the game has completely changed. We’re diving into learning-driven approaches: RL, IL, foundation models, and hybrid methods blending model-based and learning-based techniques. The advancements in humanoid locomotion and manipulation over the last decade feel like something straight out of sci-fi. 🤖✨
This is such an exciting time to be part of humanoid research. Whether you’re a seasoned researcher exploring new directions or a newcomer eager to jump in, the opportunities are endless. The field is expanding faster than ever, and there’s so much to learn, build, and contribute!
We’ve captured these trends and insights in our latest survey. 18 experts. 15 institutions. 6 months of collaboration. 🤝 It’s a true team effort, bringing together some of the brightest minds to showcase where humanoids are headed.
If you’re curious about the field or looking for inspiration, check it out! Would love your feedback—let’s keep pushing the boundaries of what humanoids can do! 🚀🤖
ArXiv: https://t.co/O5bkXK1qQq
I am launching Duke Robot Dexterity Lab (DexLab) with a focus on human-level dexterity in robotic manipulation.
The DexLab is looking for PhD students passionate about robotic manipulation starting Fall 2025!