One capture stack. Three parts.

6thSense captures touch-aware human demonstrations for dexterous robots. The glove, the skin, and the egocentric rig record together on one clock — so every episode ships as aligned, model-ready data instead of raw sensor folders.

Built for contact-rich manipulation

Most demonstration data misses the moment of contact — onset timing, pressure evolution, and the small grip adjustments that decide whether a task succeeds. The 6thSense stack is designed to capture exactly those signals, calibrated and time-aligned across every modality.

How the stack works

Capture, sync, calibrate, package — one pipeline, four stages.

Explore the stack

Gloves

Wearable tactile glove — high-rate contact and pressure streams with per-channel calibration and grasp-phase tracking.

Skin

Contact-sensing tactile skin — pressure proxies placed where touch matters for the task, with stated calibration boundaries.

Rig

Egocentric capture rig — first-person RGB, depth, and IMU aligned to the same timebase as touch.