SAMPLE RECORDS
See the data before you ask for it.
Every gate value, every aligned layer, every retargeting result — published. Files are gated; the evidence is not.
LIVE DATASET EXPLORER · SELECT A PHYSICAL LAYERFILTERS UPDATE THE SAMPLE FIELD IN REAL TIME
LAYER MATCH
Food Preparation
84 / 100
Dough shaping and proofing tray transfer
HAND + CONTACTLIVE TRACE
handcontact6DoFMOTION VECTOR
hand78%reproj2.1 pxdrift0.4 ms / 60 s
Open record
LAYER MATCH
Industrial & Production
91 / 100
Connector seating and torque check on sub-assembly
6DoF + IMULIVE TRACE
imu6DoFactionTORQUE EVENT
hand86%reproj1.8 pxdrift0.3 ms / 60 s
Open record
LAYER MATCH
Handcraft & Fine Assembly
79 / 100
Precision insertion with tweezers under magnification
HAND POSELIVE TRACE
handcontactposeMICRO-MOTION
hand92%reproj2.6 pxdrift0.5 ms / 60 s
Open record
LAYER MATCH
Specialised Services
76 / 100
Appliance panel removal and filter replacement
DEPTH + 4DLIVE TRACE
depth4D6DoFOBJECT TRACK
hand71%reproj2.4 pxdrift0.6 ms / 60 s
Open record
LAYER MATCH
General Retail
81 / 100
Shelf replenishment and facing across mixed SKUs
HOI GRAPHLIVE TRACE
handHOIactionOBJECT FLOW
hand64%reproj2.2 pxdrift0.4 ms / 60 s
Open record
LAYER MATCH
Domestic & Care Services
77 / 100
Laundry folding and sorting into categories
ACTION + POSELIVE TRACE
poseactionretargetFOLD TRAJECTORY
hand83%reproj2.8 pxdrift0.5 ms / 60 s
Open record
WHAT A RECORD SHOWS
Ingest gate values
Hand visibility, IMU-to-video drift, reprojection error and composite score — the four numbers that decide whether a clip is shipped or rejected.
Five aligned layers
The same take carrying control annotation, 4D ground truth, motion capture, 3D assets and retargeted trajectories on one timeline.
Retargeting feasibility
Pass or reject per end effector, with the reason. Rejections are retained, not discarded.
Rights chain
Which employer signed, the consent identifier, the collection region and the redaction applied.
Representative records drawn from the delivered corpus. Figures are per-clip ingest output, not corpus averages.