General Intelligence Labs

The perception-action stack for robotics.

[General Intelligence Labs @gi_labs]

Introducing Ego1, our first egocentric capture headset for Physical AI.

We co-designed the hardware and the perception stack to turn everyday first-person activity, especially manipulation, into training data for robotic models at planet scale.

[General Intelligence Labs @gi_labs]

Try Ego1:
- Synced exposure
- Hardware synced clocks
- Per device calibration

[Guanya Shi @GuanyaShi]

High-quality motion reference data is key for humanoid skill learning 🤖🕺💃

A natural idea is to leverage human motions and “translate” them to humanoid motions, a process known as retargeting. For interaction-rich tasks such as scene interaction and loco-manipulation, retargeting is challenging: it must ensure motion consistency, smoothness, kinematic feasibility (no artifacts like penetration or foot skating), and scalability (one framework can handle thousands of motions).

Excited to release OmniRetarget — a scalable retargeting method with a 4-hour high-quality humanoid motion dataset for interaction-rich tasks. OmniRetarget takes an interaction-preserving perspective: we optimize Laplacian deformation between source and target interaction meshes while enforcing kinematic constraints, producing consistent, smooth, and feasible trajectories at scale. Even better, OmniRetarget can efficiently augment motions by varying terrains, objects, and initial poses.

This high-quality interaction-preserving retargeting enables a minimal RL setup to execute long-horizon (up to 30s) agile, interaction-rich skills. All tasks in the video share just 5 rewards, 4 domain randomization terms, and rely only on proprioception.

More details: https://t.co/4lfqT126MY

[General Intelligence Labs @gi_labs]

We certainly agree!

[Chris Paxton @chris_j_paxton]

The most common misunderstanding people have when getting into robotics, is that data is useful by default. *good* data is the real fundamental limiter. and with good data (from things like omniretarget) you can do really cool stuff

[General Intelligence Labs @gi_labs]

Try Ego1 at ICRA 👀

[Shumo Chu @shumochu]

I bring Ego1 with me at ICRA. DM me if you want to try it in person.

[Javier Martin @liderarmente]

Ahora gracias a @gi_labs esto es más fácil que nunca https://t.co/oPvx0QJgyw

[General Intelligence Labs @gi_labs]

@avizurlo Thanks, Avi!

[General Intelligence Labs @gi_labs]

Introducing Ego1, our first egocentric capture headset for Physical AI.

We co-designed the hardware and the perception stack to turn everyday first-person activity, especially manipulation, into training data for robotic models at planet scale.

[Haichen Shen @shenhaichen]

Ego1 is our all-in-one egocentric headset for robot-learning data.

It captures high-quality, hardware-synced stereo video + 200Hz IMU, then turns everyday first-person manipulation into policy-training data: metric head trajectory, 3D hand tracking, and dense depth.

[Shumo Chu @shumochu]

Database system OG @YingjunWu wears an early version of @gi_labs Ego1 engineering prototype. Dumplings are yummier if you get your motion captured.

[General Intelligence Labs @gi_labs]

See the full write-up here: https://t.co/s0BKfmOsuu

DM us if you want to book a demo!

[General Intelligence Labs @gi_labs]

And dense depth for the whole scene. Stereo matching (SGBM) estimates disparity between the rectified views and converts it to metric depth using the calibrated baseline. Each frame becomes a 16-bit-mm depth map that drops into ROS, LeRobot, or Open3D.

[Shumo Chu @shumochu]

Data Collector uses our @gi_labs UMI grippers to do welding at a construction site ( video credit: @GeorgiZlatarev )

[General Intelligence Labs @gi_labs]

@DanielEdrisian We are a SF based company building data capture hardware for training models. Love to chat!

[General Intelligence Labs @gi_labs]

Robotics is increasingly a data problem. There is no lack of robots that can backflip. There is a lack of Robot who can cook Kung Bao Chicken for you. We talked to 20+ researchers, hardware vendors, surveyed 12+ research and deployed systems to write this guide. Hope you enjoy!

[Shumo Chu @shumochu]

https://t.co/qNA7W7MGoP

[General Intelligence Labs @gi_labs]

👀

[Shumo Chu @shumochu]

It cannot be more based that @UnitreeRobotics put their altruistic engineers in charge of business.

[Adam Raudonis @adamraudonis]

The robot data ecosystem is growing! Anyone I miss?
@builddotai
@DeepReach_AI
@foxglove
@GenrobotAI
@gi_labs
https://t.co/izY9e1sX0M
@labelbox
@MeckaAI
@micro1_ai
@microagi
@Neuracore_AI
@Nominal_io
https://t.co/RwPYdX6DfR
https://t.co/IftTev7XaT
@rerundotio
@roboto_ai
@ropedia_ai
@scale_AI
https://t.co/Bq43xwkB9m
https://t.co/tqxLzITISR
T* (stealth)
https://t.co/uM3OnF3fLY