RoboNetBook a demo

The Internet of Robotics

We make data collection cheaper, so you can focus on training robots.

Book a demo
$110.68per capture unit
276.5 ghand-worn device
5+grippers supported
1-2 wkpilot turnaround

01 / Problem

Robotics is data-constrained.

Training a manipulation policy takes 300 to 1,200 high-quality demonstrations per task. Most teams still collect them by teleoperating a robot, at over $100 per usable hour.

General-purpose robots require general-purpose human data.
  • Every teleop demo occupies a robot that should be deploying and validating.
  • Teleop rigs feel unnatural for bimanual and tool-use tasks, so quality drops where policies need it most.
  • Raw recordings are not a dataset. Demos need synchronized signals, labels, and QA.
  • Variation makes policies generalize, and variation is what robot-bound collection makes expensive.

02 / Solution

Collect from humans. Keep robots for validation.

Human demonstrations are several times more sample-efficient than teleoperation, and a few hundred well-curated demos now beat thousands of mediocre ones. The winning move is better data, collected off-robot.

RoboNet captures demonstrations with HandUMI, collects with trained operators in Peru, and delivers QA-reviewed, policy-ready datasets.

HandUMI Collection

Open-source hand-worn capture for bimanual arms with parallel-jaw grippers.

Capture layer

Dataset Pilots

Start with one task and validate signal quality, labels, and training usefulness.

Pilot dataset

Peru Data Operations

Lower-cost collection nodes with standardized protocols, QA, and versioned releases.

Versioned batches

03 / Why HandUMI

HandUMI is the first node in the RoboNet data network.

HandUMI is a hand-worn, open-source data collection device for bimanual arms with parallel-jaw grippers, where the operator performs the task directly and uses their own fingers as the actuator.

  • 276.5 grams
  • $110.68 per unit
  • Encoder-precision gripper aperture
  • Integrated wrist camera
  • Tracking with the VR headset of your choice: Pico or Quest
  • More than 5 grippers supported: Piper, Trossen, ARX, Soft gripper, and Dream gripper
View HandUMI on GitHub
HANDUMI / DATA_NODE
WS://CONNECTED
CAMERAS
2 wrists
APERTURE
Encoder
TRACKING
VR headset

04 / How it works

Start with one task. Scale after the data proves useful.

  1. 01

    Define the task.

    Tell us the manipulation behavior your robot needs to improve.

    Task brief
  2. 02

    Design the capture protocol.

    We define objects, setup, success criteria, labels, metadata, and delivery format.

    Capture protocol
  3. 03

    Collect demonstrations.

    Trained operators collect demonstrations through the HandUMI workflow.

    Session capture
  4. 04

    QA and package the dataset.

    We review signals, labels, metadata, and package a versioned dataset.

    QA report
  5. 05

    Evaluate and expand.

    Your team evaluates the pilot, then we expand if the data is useful.

    Versioned dataset

05 / Final vision

Full-body capture. Any embodiment.

The near-term wedge is bimanual manipulation data. The long-term RoboNet vision is a distributed network for full-body human data that can train humanoids, bimanual arms, mobile manipulators, and future robot embodiments.

  • One data network, many robot embodiments.
  • Embodiment-aware retargeting from human motion to robot policies.
  • Standardized capture rigs, schema, validation, and long-term expansion.
Plan a data collection pilot
RoboNet full-body capture vision mapped from human demonstrations to multiple robot embodiments

06 / Use cases

Built for the manipulation tasks where generic data falls short.

Humanoid robotics

Collect human demonstrations for household, workplace, bimanual, and tool-use tasks that humanoid policies need to understand.

BIMANUALTOOL_USEVIDEO

Robot foundation models

Build diverse human hand/object interaction datasets for embodied AI systems that need broad manipulation priors.

OBJECT_SETVARIATIONQA

Academic robot learning labs

Run focused data pilots without building a full internal data collection operation for every paper or benchmark.

PILOTSCHEMARESEARCH

Warehouse automation

Capture picking, packing, sorting, bag handling, container loading, and long-tail SKU interaction data.

PICK_PACKSKUEDGE_CASES

Food robotics

Collect demonstrations for deformable, fragile, irregular, and tool-mediated food handling workflows.

DEFORMABLETOOL_USEFAILURES

Lab automation

Capture pipette-adjacent workflows, vial handling, rack interaction, tray loading, labware transfer, and repetitive bench tasks.

LABWAREREPEATABLEMETADATA

Bimanual manipulation

Record coordinated two-hand demonstrations for tasks that cannot be reduced to single-arm grasping.

TWO_HANDWRIST_POSEAPERTURE

FAQ

Common pilot questions.

RoboNet collects task-specific manipulation datasets for robot learning teams using HandUMI and QA-reviewed workflows.

The main offer is data collection. HandUMI is the capture layer that enables the workflow.

Human demonstrations are more sample-efficient than teleoperation for the same collection time, and they capture natural dexterity that teleop rigs degrade. Robot hardware is then free for validation and embodiment-specific tuning.

RoboNet combines Peru-based operations, lower facility costs, and lower-cost HandUMI hardware.

No. The model is built around cost-efficient quality: SOPs, calibration, task protocols, operator training, metadata, labels, QA review, and dataset acceptance criteria.

Yes. The recommended starting point is one task, one capture protocol, and a focused pilot dataset your team can evaluate before scaling.