UMI

appreciate @pronounced_kyle's framing of "read" "write" access between digital and physical worlds. One of the fundamental challenges for automating manufacturing is to solve for the digital-to-physical "write" access and…

Ekin Dogus Cubuk of @periodiclabs on why AI has to learn from the physical world, and building their own lab to close that loop: "One of the more fundamental ways of making things around us is powder synthesis... You can actually, using…

AI models are getting close to designing hardware better than humans, and that shifts where the hard problems are: - Better simulation, so physical builds become confirmation instead of experiments - Demand shifts toward parts, machining…

"It took me three months to design it, but it's nine months until I can actually start to use it." Former Intel CEO Pat Gelsinger on the manufacturing bottleneck between designing hardware with AI and actually getting it built: "Let's say…

For AI to become a superhuman hardware engineer, it has to learn from the physical world, and real-world data is hard to get: - It's non-uniform and constantly changing - It's sparse: you can't measure everything all the time - It's…

AI is on track to become a superhuman hardware designer, but design is only the beginning. The interface between bits and atoms is lossy, and turning designs into real hardware gets harder at every level: 1. Sourcing off-the-shelf parts…

『拡張性の高い人体操作データ収集用外骨格』 https://t.co/WOKWHxFjii #exoskeleton #manipulation #DataCollection #PhysicalAI #UMI #interface #WearableDevice #AeroHand #ChestnutRobotic https://t.co/qCm1E6HlIw

I really want to know more about how their stack works because the videos are consistently mind blowing. very very fast and smooth, humanlike motion. Fairly sure this is UMI-style work, not teleop data. https://t.co/Jjib57wqJ1

Collecting good data is extremely important for robotics right now, and i think UMI seems most promising https://t.co/jslSvX7SGR

UMI, but for humanoid hands. UMI-style capture already works great for simple grippers. For a high-DOF hand, where the action space is huge and contact force carries the important signal, matching the data collector to the hand matters…

Robotic data sits at the heart of robotics. Four ways to collect it + open datasets: ✸ Teleop | ABC-130K | 3.5k hours ✸ UMI | HiFi-UMI-2K | 2k hours ✸ Egocentric | Egocentric-10K | 10k hours ✸ Synthetic | Genie Sim 3.0 | 10k+ hours…

Post Log in Sign up Post huaijiang on X: "ok, so if i understood this correctly, the “in-context learning” ability was essentially baked into the model at training time. roughly speaking, the policy is trained to do something like: (demo…

the most interesting hardware in robotics isn't robots. @sundayrobotics built a skill capture glove so regular people can record chores. proception built a glove for dexterous hand data. UMI turned gripper data collection into a $300 tool.…

Overview video Figure 1: Action flow as a shared control interface. Top: Hydra-0 learns from diverse interaction videos featuring egocentric human demonstrations, handheld UMI grippers, bimanual robot arms, and single-arm robots. Middle:…