Grabette Slam Test
Convert raw Grabette recordings to LeRobot dataset
Robot learning has a supply problem. We have capable policy architectures (transformer-based VLAs, diffusion and flow-matching policies, and even world models) and the GPUs to train them. What we lack is large, diverse, real-world manipulation data.
Teleoperating a robot to collect it can be expensive and demanding: first of all, it requires a robot. And depending on the teleoperation method, data collection can be tedious for the user if it takes hours and involve significant hardware and logistical challenges. That is difficult to scale with the wide variety of tasks and environnements required.
But you don't need a robot to collect robot data. Just a human hand, a gripper, a camera, and a way to recover the 6-DoF trajectory of what the hand did. Capture the demonstration and you have data a robot can learn from.
That's what we're releasing today: Grabette, an open, low-cost system for recording manipulation data. Pick it up, record a task with your own hand, and get back a clean, robot-ready dataset. No robot, no lab, no teleop rig.
And that's the bigger goal: if recording a demonstration is as easy as shooting a video, anyone can contribute. We want Grabette to seed a large, open, collaborative manipulation dataset. One no single lab could ever build alone.
Grabette is directly inspired by the Universal Manipulation Interface (UMI) from Stanford: a handheld gripper with a fisheye camera that records demonstrations "in the wild", recovers camera trajectories with SLAM, and trains visuomotor policies from them.
UMI proved the recipe works. Other (closed source) devices exists like Agibot's MEgo gripper, Genrobot's DAS gripper and Sunday Robotics skill capture glove. Our goal was to make it effortless to use, to get the barrier from "I have a task" to "I have a trained model" as low as possible.
Grabette is built into the modern open ecosystem: LeRobot for datasets, the Hugging Face Hub for sharing, and a processing pipeline you run from your browser with nothing to install. Grabette is something anyone can build on a workbench, use in the field, and contribute data from.
We have been developing Grabette for months, and we feel it has become usable enough to share. We are excited to share it now with you!
Grabette is a handheld gripper instrumented with everything needed to reconstruct a manipulation demonstration.
It carries two cameras, each with a distinct job. Splitting the two roles is deliberate: the cheap wide fisheye gives the policy the context-rich, wrist-camera-style view it needs, while the RGBD camera does the heavy lifting of robust 6-DoF tracking.
And while Grabette records data during tasks performed by a user, it relies on its robotic counterpart to execute the movements it has learned after training. So, meet Gripette, the robotic arm end-effector twin of Grabette.
The family shares the same hardware DNA:
Grabette hand held recording device (left) and Gripette robot gripper (right)
Everything is open source
Components. Standard sensors you can buy, no closed pipeline, no fork lock-in. A Raspberry Pi, a standard Pi camera, an off-the-shelf OAK-D depth camera, magnetic encoders. The whole point is that anyone can build one from parts you can just order.
Robot-agnostic by design. Nothing in the capture or the data format assumes a particular arm. Demonstrations are stored as camera-local 6-DoF cartesian pose plus gripper state, the output is a standard LeRobot dataset on the Hugging Face Hub, so the same data can drive different robots and different learning methods. You will still need the matching Gripette gripper on your arm though.
This release enables anyone to go from "I want to demonstrate a task" to "I have a training-ready dataset" quickly and without prior expertise.
Press the button, and data from the observation camera, the tracking camera (color, depth, and IMU), and the gripper’s encoder joint values are recorded simultaneously, using a single shared clock to ensure proper synchronization. Press the button again to stop the episode, and the data is saved locally on the Raspberry Pi.
Open the Grabette dashboard in your browser. Select the episodes you want to add to the dataset, and with one click, post-processing begins.
Everything is now ready to start training!
A dataset is only interesting if it trains something. So, to show the loop end-to-end, we ship a complete example. But to be clear, this is an example of what is enables, not the product: the release is the recording system. Your Grabette data works with any method that consumes LeRobot datasets.
Our example takes 200 recorded demonstrations such as :
Dataset available on this HF Dataset
and:
Policy available on this HF Model
The data bottleneck doesn’t get solved by one lab, but by a community recording demonstrations everywhere. You can now help build the dataset: build a Grabette, record tasks you are interested in, and share them on the Hub. Every episode makes the open dataset bigger and more diverse, and every contributor makes robot learning a little less gated behind expensive hardware.
This release is only the start of the project! Grabette will keep evolving. We already have more coming, including Casquette, a head-mounted POV device to complement Grabette for egocentric capture (still a work in progress). But the most important next step isn’t only ours, it’s also yours: start recording, and let’s build the dataset together.
👉 Build a Grabette (GitHub) · Process your data (HF Space) · Contribute to the dataset
Built with ❤️ by Pollen Robotics.
Convert raw Grabette recordings to LeRobot dataset
Facts Only
* Pollen Robotics developed Grabette and Gripette.
* Grabette is a handheld recording device featuring a Raspberry Pi, a Pi camera, an OAK-D depth camera, and magnetic encoders.
* Gripette is a robotic arm end-effector twin of the Grabette.
* The system records observation camera data, tracking camera data (color, depth, IMU), and gripper encoder joint values.
* Data is processed through a browser-based pipeline and stored as LeRobot datasets.
* Datasets are shared via the Hugging Face Hub.
* The hardware is open source.
* Grabette is inspired by Stanford's Universal Manipulation Interface (UMI).
* Comparable closed-source devices include Agibot's MEgo, Genrobot's DAS, and Sunday Robotics' skill capture glove.
* An example policy was trained using 200 recorded demonstrations.
* A head-mounted POV device called Casquette is currently under development.
Executive Summary
Pollen Robotics has released Grabette, an open-source, handheld recording system designed to collect real-world manipulation data for robot learning without requiring a physical robot during the capture phase. By using a handheld gripper equipped with a wide-angle fisheye camera for context and an RGBD camera for 6-DoF tracking, users can record human demonstrations that are then processed into LeRobot-compatible datasets and shared via the Hugging Face Hub.
The system is designed to lower the barrier to entry for creating diverse manipulation datasets, which currently face a supply bottleneck due to the high cost and complexity of traditional teleoperation. To complete the loop from data collection to execution, Pollen Robotics also introduces Gripette, a robotic end-effector twin that allows a robot arm to execute the learned movements. While the system aims to seed a collaborative, open-access dataset, its effectiveness depends on community adoption and the hardware requirement of using the matching Gripette gripper on the target robot arm.
Full Take
The strongest version of this narrative is that the "data bottleneck" in robotics can be solved through the democratization of data collection, shifting the burden from expensive labs to a distributed community of contributors using low-cost, open-source hardware.
This is a classic "platform play" disguised as an open-source liberation movement. While the hardware is open and the data is shared, the utility of the data is tied to a specific ecosystem (LeRobot) and a specific hardware requirement (the Gripette gripper). The narrative frames the current state of robotics as "gated" by expensive hardware, positioning Grabette as the key to unlocking it. However, the reliance on a specific proprietary-style gripper twin for execution creates a new, albeit lower, point of lock-in.
Patterns detected: ARC-0073 Predatory "liberation" rhetoric, ARC-0012 Authority Game
The driving paradigm is the "Large Model" philosophy: the belief that scale and diversity of data are the primary drivers of intelligence, regardless of how that data is sourced. It assumes that human-led demonstrations recorded via a handheld tool are a sufficient proxy for robot-led manipulation.
The benefit accrues to the broader AI research community and Pollen Robotics, who position themselves as the curators of a critical infrastructure (the dataset). The cost is borne by the community contributors who provide the labor of recording.
Bridge Questions:
1. To what extent does the "human-to-robot" transfer fail when accounting for differences in kinematics between a human wrist and a robotic joint?
2. Does the requirement for a matching Gripette gripper negate the "robot-agnostic" claim of the data format?
Counterstrike Scan: A coordinated campaign would use the "open source" label to mask the creation of a new industry standard that forces competitors to adopt their hardware ecosystem to stay compatible with the emerging "community" dataset. The content aligns partially with this pattern by emphasizing liberation while maintaining a hardware dependency for execution.
Sentinel — Human
This text reads like an enthusiastic project announcement from a development team, blending detailed technical specifications with a strong vision for open-source community collaboration.