6thSense
Products

One system, three instruments.

The glove, the robot skin, and the ego rig were built together. Sync and calibration live in the hardware, not in a script you maintain.

01

Capture glove

Tactile · hand pose · IMU

Worn by the demonstrator. Records pressure, contact timing, hand pose, and motion through the task.

96 channels
Pressure across palm, finger pads, and lateral surfaces
1 kHz
Tactile sample rate
1–300 kPa
Calibrated per channel, ±4% repeatability
26-DoF
Hand pose and grasp phase at 120 Hz
9-axis
IMU at 400 Hz
165–205 mm
Hand lengths across three shell sizes
90 s
Calibration before each session
02

Robot skin

Allegro · LEAP · Shadow

The same sensing stack molded 1:1 to a robot hand, so what you capture on human hands transfers.

1:1
Molded to your hand’s geometry
Allegro · LEAP · Shadow
Fitted today
CAD
Send it for any other hand
Same semantics
Channel definitions match the glove
03

Egocentric rig

Egocentric RGB · depth

The head rig. First-person video and depth, timestamped in hardware against the glove.

1080p60
First-person RGB with exposure locking
640×480
Depth at 30 Hz on the ego timebase
<2 ms
Held against the glove across a full session
Q2 2027
Wrist and scene viewpoints

What the rig records.

Eight streams, one timebase, per-channel semantics you can actually train against. Packaged per episode for LeRobot, RLDS, or HDF5.

8 × 12 · 96 channelsHover · hold to press
ch ---0 kPa1 kHz · 1–300 kPa
1 kHzTactile & pressure proxiesGlove
96 channels at 1 kHz across palm, finger pads, and lateral surfaces. Calibrated per channel over 1–300 kPa.
1080p60Egocentric videoEgo rig
First-person RGB at 1080p60, with exposure locking for long household sessions.
30 HzDepth (RGB-D)Ego rig
640×480 at 30 Hz on the ego timebase, for geometry, reach, and clutter around the hands.
120 HzHand poseGlove
26-DoF articulated hand state and grasp phase at 120 Hz. Not 2D boxes in frame.
400 HzMotion & IMUGlove
9-axis at 400 Hz: inertia, rhythm, and effort through the task.
Q2 ’27Wrist & scene camerasEgo rig
Secondary viewpoints for occlusion recovery and tool use. Shipping Q2 2027.
AlignedLabels & commentaryCapture unit
Task and subtask boundaries, contact phases, QC flags, plus optional frame-aligned text narration.
GradedSuccess / failure outcomesCapture unit
Binary or graded success, failure modes, and segment-level tags for imitation and eval.

One glove, one fitted skin, two weeks of capture.

No long-term commitment. You own every episode.