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

Industry·9 min read·August 14, 2026

Who Builds Robot Manipulation: A Registry of 39 Labs and Companies

A cleaned, sourced registry of the academic labs and companies building robot manipulation and the data behind it, from Stanford IRIS and Berkeley RAIL to Physical Intelligence, Figure, and AgiBot. One schema for the whole field.

The people building robot manipulation are scattered across university lab pages, company blogs, and press releases, and there is no single place to see them side by side. We refined thirty-nine of them, twenty academic labs, nineteen companies, into one schema: who they are, who leads them, what they work on, what hardware they use, and the notable dataset or model they produced. The machine-readable version is at blomegalab.com/data/robotics-labs.json.

An academic frontier and an industrial buyer base

The registry splits cleanly. Twenty entries are academic labs, the places where the datasets and methods are invented: Stanford IRIS and SVL, Berkeley RAIL and AUTOLAB, CMU R-PAD, MIT Improbable AI and the GelSight group, UT Austin RPL, Shuran Song's REAL, and strong groups at Bristol, ETH, Imperial, Oxford, SJTU, and Tsinghua. Nineteen are companies, split between foundation-model labs (Physical Intelligence, Skild, Generalist, NVIDIA GEAR, Dyna, TRI, DeepMind) and humanoid makers (Figure, 1X, Apptronik, Sanctuary, AgiBot, Unitree, Fourier, UBTech). The field is US-led, twenty-four of thirty-nine, with China the clear second concentration at six.

39 entries: 20 academic labs, 10 humanoid companies, 9 foundation-model/data companies. Labs invent the data; companies consume it. That gap is a market.

The registry (selected)

A representative slice below; the full thirty-nine, with lead, focus, hardware, and notable dataset for each, are in the JSON.

How it was built

Each entry was pulled from a lab website, a company page, or reporting, and normalized to the same fields with a confidence level. A handful of URLs and leadership fields are marked medium confidence where the primary page had moved or redirected, and where a group has no single flagship dataset that field is left blank rather than invented. The registry is a starting map, not a census; it will grow.

Why Blomega maps this

The split in the chart is the point. Academic labs invent the datasets; companies consume them, and most of the companies build robots faster than they can collect the diverse, force-instrumented human data those robots need. The dataset registry showed the force gap in the data; this registry shows the organizations on both sides of it. It is the map of who produces manipulation data and who is short of it, which is exactly the terrain a first-party data company operates on.