---
title: "Grabette: an open system to record robot-manipulation data"
description: "Grabette is an open, low-cost system for recording robot-manipulation data without requiring a robot, lab, or teleoperation rig. It uses a handheld gripper with a wide fisheye observation camera, an RGBD tracking camera, an IMU, and gripper encoders to capture synchronized demonstrations and recover camera-local 6-DoF Cartesian pose plus gripper state. Users save episodes on a Raspberry Pi, then select them in a browser dashboard for upload, RTAB-MAP SLAM, trajectory verification, and conversion into a LeRobot dataset on the Hugging Face Hub. The release also includes a motorized Gripette gripper and an example that trains a Diffusion Policy on 200 demonstrations and evaluates it on an OpenArm 7-DoF arm via gRPC, demonstrating the recording-to-training workflow."
---

# Grabette: an open system to record robot-manipulation data

[Hugging Face](https://yomu.fyi/company/hugging-face) · Steve Nguyen, Claire Houziel, Gaelle Lannuzel, Simon Le Goff, Jeremy Laville, Étienne · Jul 21, 2026

**Type:** Announcement

## Summary

Grabette is an open, low-cost system for recording robot-manipulation data without requiring a robot, lab, or teleoperation rig. It uses a handheld gripper with a wide fisheye observation camera, an RGBD tracking camera, an IMU, and gripper encoders to capture synchronized demonstrations and recover camera-local 6-DoF Cartesian pose plus gripper state. Users save episodes on a Raspberry Pi, then select them in a browser dashboard for upload, RTAB-MAP SLAM, trajectory verification, and conversion into a LeRobot dataset on the Hugging Face Hub. The release also includes a motorized Gripette gripper and an example that trains a Diffusion Policy on 200 demonstrations and evaluates it on an OpenArm 7-DoF arm via gRPC, demonstrating the recording-to-training workflow.

## Context

Robot learning is constrained by the shortage of large, diverse, real-world manipulation data. Conventional teleoperation can require expensive hardware and make data collection tedious and difficult to scale across tasks and environments.

## Approach / What changed

Grabette captures demonstrations with a handheld gripper using synchronized observation, RGBD tracking, IMU, and encoder data. A Raspberry Pi stores episodes, while a browser-based pipeline uploads them, runs RTAB-MAP SLAM and trajectory checks, and converts them into standard LeRobot datasets. The accompanying Gripette motorized gripper supports execution on a real or simulated robot arm.

## Takeaways

- Grabette records camera-local 6-DoF Cartesian pose and gripper state in a robot-agnostic format that is converted to standard LeRobot datasets.
- The capture workflow combines a wide fisheye camera for policy observations with an RGBD camera for 6-DoF tracking, plus synchronized IMU and encoder data.
- An included example trains a Diffusion Policy from 200 demonstrations and evaluates it on an OpenArm 7-DoF arm using the Gripette gripper and a gRPC API.

**Tags:** [Data Pipelines](https://yomu.fyi/topic/data-pipelines), [gRPC](https://yomu.fyi/topic/grpc), [Machine Learning](https://yomu.fyi/topic/machine-learning), [Open Source](https://yomu.fyi/topic/open-source)

- Source: [Hugging Face](https://huggingface.co/blog/grabette)
- Source URL: https://huggingface.co/blog/grabette
- Ingested by Yomu: 2026-08-29T05:40:20.783Z

[Read original post](https://huggingface.co/blog/grabette)