Training data for the tasks robots can't do yet

We capture what cameras miss — force, contact, recovery — and deliver physics-verified episodes ready for your training pipeline.

Syngraph Visualization

FROM CAPTURE TO DEPLOYMENT

Every episode passes through four stages before delivery

01

CAPTURE

Multi-camera egocentric video, LiDAR depth, and tactile force sensing. Skilled workers performing real manipulation tasks. Synchronized to sub-10ms.

02

ANNOTATE

Automated pre-labeling — hand pose estimation, object segmentation, action classification — on every frame. Expert human review on the edge cases.

03

VALIDATE

Physics validation gate. Force ranges, joint limits, contact sequences checked against physical constraints. Failures and recoveries included by design.

04

DELIVER

LeRobot v3 format. GR00T N1 compatible. Drop it into your training pipeline. Delivered via bucket or direct transfer.

BUILT FOR

FOUNDATION MODEL TEAMS

Pretraining at Scale

Diverse scenes, camera poses, and task types for VLA and world model pretraining — with tactile and force data no open dataset provides.

DEPLOYMENT TEAMS

Contact-Rich Precision

Task-specific episodes with synchronized tactile and depth. Failure cases and recovery behaviors. The contact-rich signal your sim-trained policy is missing.

4

Synchronized modalities per episode

LeRobot v3

Native format

YES

Failure + recovery included

GR00T

N1 compatible

FAQ

LeRobot v3 (synchronized MP4 video + Parquet state/action files). Compatible with GR00T N1, Isaac Lab, and standard VLA training pipelines. Delivered via bucket or direct transfer.

Egocentric RGB video (multi-camera), LiDAR depth maps, 3-axis tactile force and shear sensing, 3D hand kinematics (MANO format), and language-annotated task descriptions. All synchronized to sub-10ms.

Every episode passes a validation gate before delivery. Force values checked against biomechanical norms. Joint angles verified against anatomical limits. Contact sequences validated for physical consistency. Episodes that fail are quarantined, not shipped.

Precision assembly, tool use, textile manipulation, surface quality inspection, and more. Each domain captures contact-rich, bimanual tasks where vision-only systems fail. See our task library.

Yes. Our capture protocols are designed to include failures and recovery behaviors alongside successful demonstrations. Models need to learn what goes wrong and how to correct, not just the golden path.

Contact us. Tell us what task domains you need and what's missing from your current training corpus. We'll scope it from there.

SEE IT IN YOUR PIPELINE

Tell us what task domains you need and what's missing from your training corpus. We'll scope it from there.