data-collection-ui-preview
OPEN-SOURCE RESOURCES

FRANKA LABS

LABS (Learning & Acquisition of Behaviors Suite) is an open-source, microservice-driven solution for robotics teleoperation and data collection. It is designed around Franka reference setups and can be extended to your own Franka-based cells or additional robots and peripherals through its modular architecture.


ABOUT LABS

  1. FOCUS REMAINS ON RESEARCH, NOT ON INTEGRATION

    Researchers should focus on teleoperation, data collection, training, policies deployment, and benchmarking. LABS removes the integration boilerplate across these stages, reducing engineering overhead and enabling faster iteration on research.

  2. CONSISTENT WORKFLOWS FROM COLLECTION TO DEPLOYMENT

    Using shared drivers, controllers, and interfaces across the full pipeline ensures consistency between data collection, evaluation, and deployment, improving reproducibility and reducing duplication.

  3. READY-TO-USE WITH REFERENCE SETUPS

    LABS provides reference implementations for FR3 (Droid setup), FR3 Duo (bimanual reference setup), and Mobile FR3 Duo (mobile bimanual reference setup). These configurations can be used directly or adapted to custom Franka-based cells or research setups.

  4. MODULAR AND EXTENSIBLE ARCHITECTURE

    All components run as independent containerized services, including robot drivers, teleoperation devices, sensors, recording, and processing. The system can be extended or modified without changing the core stack.

WHAT IS INCLUDED
_Frank Robotics0363 (1)-1

THREE PIPELINE SERVICES, ONE STACK

 

Data Collection
A FastAPI-based orchestration service that manages the workflow state machine, including teleoperation mode, task selection, and episode lifecycle control through a web UI.

 

Data Recorder
A ROS 2 recording service that captures synchronized topics from all active devices, including robots, sensors, and teleoperation inputs. Data is stored as MCAP episodes.

 

Data Processor
A post-processing service that transforms recorded episodes, applies video encoding, and exports datasets into LeRobot v2.1 format for downstream training workflows.

LABS

FROM SETUP TO DATASET IN FIVE STEPS

LABS is designed to minimize time between hardware setup and the first recorded episode.

  1. 1. CONFIGURE YOUR STATION

    Define your hardware setup in a deployment configuration, including robot IPs, sensor types, and device mappings.

  2. 3. COLLECT DEMONSTRATIONS

    Open the Data Collection UI, select a task, start teleoperation, and record episodes.

  3. 2. START THE SOFTWARE STACK

    Deploy the system using Docker Compose and Tilt. The stack runs as a set of containerized microservices managed through a live development environment.

  4. 4. RECORD AND STORE DATA

    Episodes are automatically recorded and stored in MCAP format with synchronized sensor and robot streams.

  5. Screenshot 2026-06-01 at 17.31.32

  6. 5. EXPORT DATASET

    Run the data processor to convert recorded episodes into a structured LeRobot v2.1 dataset.

PROTOTYPE PRODUCTS
260218_Camera01_fp (1)_resized

THE REFERENCE PLATFORM FOR PHYSICAL AI

A new generation of robot systems designed to enable physical intelligence research, from tactile teleoperation to mobile manipulation. These systems enable physical AI activities, including data collection at scale and real-world learning.

 

DEVELOPMENT
Franka Research 3 with Ai Companion integration
BUILD, CREATE, INNOVATE

DEVELOP ON THE FRANKA ROBOT PLATFORM

The resource hub for developers working with the Franka Research 3 robot platform. Dive into a wealth of tools, libraries and documentation designed to accelerate your development journey. Whether you're a seasoned developer or just starting out, you'll find everything you need to build, create and innovate. 

FOLLOW US