Digitus Labs · Embodied AI Infrastructure
Data Infrastructure Layer
for Embodied AI
High-fidelity, sub-10 ms latency joint-angle and force telemetry for humanoid robotics labs.
Digitus Labs Inc
Training data for humanoid manipulation
Human hands, machine-readable.
Digitus Labs builds position and pressure mapping gloves. Every joint angle and every newton of grip is recorded straight off your hand. The firmware exports episodes in the format your training stack can read, avoiding occlusion, increasing accuracy, and making training data more clean, accurate, and precise.
The Glove
DG-1 capture glove
0.1-100 N
calibrated fingertip pressure, normal + shear
26 DoF @ 120 Hz
every finger joint, measured on the hand
< 10 ms
force and position on one clock, verified per episode
under 2 min
put the glove on and start recording in under two minutes per session
Comparison
What each kind of data actually captures
Every way robots learn manipulation today, side by side. Only one of them measures force.
| Data source | Finger position | Grip force | During contact | Cost to scale |
|---|---|---|---|---|
| ◐Internet & egocentric video | Guessed from pixels | None | Accuracy halves | Cheap, scrapable |
| ◐Vision hand-tracking | Estimated | None | Fails at the grasp | Cheap |
| ◐Robot teleoperation | Robot joints | Rare, uncalibrated | Yes | A robot per operator |
| ◐Simulation | Perfect but synthetic | No validated contact physics | Yes | Near-free |
| ●Digitus glove | Measured · 26 DoF @ 120 Hz | Calibrated · 0.1-100 N | Unaffected | Human-speed, no robot |
● measured / holds up · ◑ partial or estimated · ○ missing or breaks
Robots can see okay, but not well. And they cannot sense at all.
Humanoids learn by watching video, that is why one can pour coffee in a demo and still crush a strawberry in actual practice. It has no idea how hard it is gripping.
Touch is the other half of manipulation, and there is no internet to scrape that knowledge from. Someone has to record it from real human hands, and that is the gap we aim to address.
Language models had the entire internet to train off of, but robots have near nothing. We are building the training data for the physical world, and every glove shipped adds to it.
Software
You record. The software does the rest.
The software timestamps both streams, aligns them, and writes training episodes in the format your stack reads. There is nothing to build on your side. It works like this the day it arrives.
fingertip force · 0.1-100 N
where every joint was · 120 Hz
timestamped on the glove · <10 ms between streams, checked on every episode
force + position, one timeline · grasp + release moments marked · mapped to your robot’s hand, written in your training format, straight off the glove
No cameras, no cloud. Your data never leaves your lab.
Consumer Edition
For consumers. Labs, use the form below.
The first production run goes to research partners; a consumer edition with the same sensing comes after, for creators, VR, and telepresence. This waitlist is for consumers only. If you are a lab, use the partner form in the next section.
Partners
Tell us where your policies drop the egg.
We work directly with humanoid and foundation-model labs and with academic research groups. Tell us the hand you are training and the format you need. You will have a draft LOI and a quote within two business days.
Industry LOI
reserve gloves from the first production run at locked pricing
Academic partnership
research pricing, free pipeline software, co-authorship
Consumer edition
not yet. Join the waitlist above and we will email you first
Start an LOI
Tell us the hand you are training and the format you need.