How to Record Robot-Manipulation Data
Grabette is an open system that allows users to record robot-manipulation data using a handheld gripper, and turn it into robot-ready datasets automatically. This innovation aims to address the bottleneck of limited real-world manipulation data in robot learning. With Grabette, users can contribute to a shared dataset, promoting collaborative progress in the field.
The thing is, creating robust robot learning models is a real challenge, mainly because we're missing out on large, diverse datasets of real-world manipulation - you know, the kind of data that's hard to come by. Traditional methods, like teleoperating a robot, are not only expensive but also super demanding and tough to scale up. That's where Grabette comes in, a new system that's trying to solve this problem. It's an open, low-cost system that lets users record manipulation data using a handheld gripper, a camera, and a way to recover the 6-DoF trajectory of the hand's movements.
Grabette actually draws inspiration from the Universal Manipulation Interface (UMI) from Stanford, which uses a handheld gripper with a fisheye camera to record demos and recover camera trajectories with SLAM. The Grabette system is made up of a handheld gripper instrumented with two cameras - an RGBD camera for robust 6-DoF tracking, and a wide fisheye camera for context-rich views. What's cool is that the data collected by Grabette can be processed right in a browser and shared on the Hugging Face Hub, making it super easy for anyone to contribute to a shared dataset.
The potential of Grabette is huge - it could make recording manipulation data as simple as shooting a video. This could lead to the creation of a massive, open, collaborative manipulation dataset that no single lab could build on its own. With Grabette, users can record their own manipulation tasks in just minutes, turn them into robot-ready datasets automatically, and help grow the shared dataset. Plus, the system is robot-agnostic, so the same data can be used to drive different robots and learning methods. By making Grabette open source, the developers are hoping to encourage collaboration and progress in the field of robot learning.
Source: Hugging Face
Related posts
Texas AI Data Centers Threaten Air Quality Progress
A Houston Advanced Research Center report warns Texas's AI data centers, leading the U.S. with 26 gigawatts planned, could reverse decades of air quality progress within five years. If powered by the grid's current mix, fine particle pollution would triple, nitrogen oxides would surge 136%, and carbon emissions would jump 90%. Onsite renewables could cut nitrogen oxide emissions tenfold, but developers choose gas turbines for speed, risking public health and regulatory crackdowns.
IBM Releases Granite Speech 5.0 With Faster Multilingual Models
IBM released three updated Granite Speech models on Hugging Face October 21, 2026. The flagship Granite Speech 5.0 2B adds multilingual transcription and translation with higher accuracy. The 470M TurboCTC model now supports seven languages and achieves 19,870 RTFx, a 1.6× speedup. The 4.1 2B Plus improves speaker attribution and timestamp precision for multi-speaker recordings. All are Apache 2.0 licensed.
Walmart Opens Its Website to AI Shopping Agents Next Month
Walmart announced starting next month AI agents can shop directly on its site via an "Agent Gateway" using MCP. Agents are restricted to shopper intent, limited data, and separate profiles. Purchases require explicit approval by default, but users can set spending caps for autonomy. Sam's Club will offer the same access. This contrasts with Amazon, which reportedly blocks external agents. Beta gateway is live for developers.