BLO LAB (Blomega) builds wearable sensing hardware, the GX-1 glove and HC-1 head camera, that captures human hand skill (finger bend, fingertip force, wrist motion, and time-synced POV video) and turns it into training data for robot manipulation and embodied-AI models.

What does the GX-1 glove measure?

The GX-1 records 10 finger-bend channels (2 joints per finger), fingertip grip force on thumb, index and middle fingers, and a 6-axis wrist IMU, all at 100 Hz, hardware-synchronized and video-sync capable.

How can I get paid to capture data for BLO LAB?

Skilled people join Talika (talika.ai), wear the gear, and complete real-world tasks. Every verified hour of skilled work becomes training data and a paycheck; gear ships to your door.

All articles

Research·10 min read·August 14, 2026

The Tactile and Force Sensor Registry: 24 Ways to Give a Robot Touch

A cleaned, sourced registry of the tactile and force/torque sensors robots use, from open-source GelSight and ReSkin to industrial ATI and Bota. One normalized schema, and a split that tells you where the field is going.

Touch is the sense a robot needs most for contact-rich work and the one hardest to shop for. The sensors are scattered across vendor datasheets, arXiv papers, and lab pages, described in incompatible terms: some report camera pixels, some report taxels, some report six axes of force. We refined twenty-four of them into one schema, so you can finally compare a vision-based GelSight to a magnetic ReSkin to an industrial ATI force/torque cell on the same row. The machine-readable version is at blomegalab.com/data/tactile-force-sensors.json.

Two speeds: an open tactile frontier over a closed force base

The clearest pattern in the data is a split. Every open-source sensor in the registry is a research tactile sensor, GelSight's DIGIT, MIT's GelSlim, Bristol's TacTip, Meta and CMU's ReSkin, NYU's AnySkin and eFlesh, the low-cost 9DTact. Every industrial force/torque sensor, ATI, Robotiq, Bota, OnRobot, and every commercial tactile array, is proprietary and usually quote-only. The frontier of touch is being built in the open by labs; the reliable, calibrated force base is a closed industrial market. A robot builder lives across both.

Of 24 sensors, all 8 open-source ones are research tactile sensors and all 15 proprietary ones are industrial force/torque or commercial arrays (1 unconfirmed). Touch is being pioneered in the open; force is sold closed.

The registry

Twenty-four sensors, each normalized to the same fields: maker, sensing type, what it senses, resolution, sample rate, output, whether it is open-source, price when public, and year. The full queryable version is the JSON linked below.

How to read it

The registry spans four sensing principles: vision-based tactile sensors that watch an elastomer deform through a camera (GelSight, DIGIT, GelSlim, TacTip, 9DTact); magnetic skins that read a magnetometer under a magnetized elastomer (ReSkin, AnySkin, eFlesh, uSkin, PaXini); optical pillar arrays (Contactile); and strain-gauge or capacitive force/torque cells (ATI, Robotiq, Bota, OnRobot). The honest gaps: sample rate is missing or controller-dependent for a quarter of rows, and most industrial prices are quote-only, so both are flagged rather than guessed. Sparsh, Meta's tactile foundation model, is deliberately excluded because it is a model that runs on these sensors, not a sensor itself.

Why Blomega maps this

This is the hardware side of the force gap. The dataset registry showed that almost no training data carries force; this registry shows the instruments that produce it, and that they are either research prototypes or expensive industrial cells, not something wired into a wearable capturing real human work. That space, force sensing on the hand, during real tasks, at data-collection scale, is where Blomega's GX-1 sits. Mapping the sensor landscape is how you see the shape of the hole.