LITCHI T1 Series

Teleoperation and imitation learning data collection in one glove. Wireless and low-latency, it captures hand motion with high precision, in three models to fit your use: T1 Lite, T1 and T1 Pro (with tactile sensing)

Human hand motion, straight into robot training data.

The LITCHI T1 Series is a glove for teleoperation and data collection developed by LITCHI, a Physical AI infrastructure company with the motto “Let Intelligence Flow”. Multi-IMU pose tracking reproduces hand motion with high precision, and by syncing with the first-person (Ego) data collection device it records both “what the operator sees” and “how the hands move” at real work sites.

It captures data reliably even near motors and high-current wiring, making it well suited to teleoperation close to robots. LITCHI runs its own end-to-end collection, processing and quality control of real-task data, and the T1 is the equipment used in that work. As the authorized distributor in Japan, HumansX supports you in Japanese from product supply to collection setup and technical support.

Demo: a single T1 glove teleoperating several robot hands with different designs at the same time

Three models for different applications

T1 Lite | Lightweight modelLightweight motion capture with 7 IMUs. An entry point for teleoperation, gesture capture and imitation learning.

T1 | High-precision model11 IMUs record the fine motion of every finger. Static orientation accuracy 0.05° or better, wireless latency 2 ms. *No tactile sensors (IMU only).

T1 Pro | Full-featured IMU + tactile modelBeyond 11 IMUs, tactile sensors at 368 points across the palm (expandable to 600) record where and how hard the hand touches.

If you need tactile or contact-force data, choose T1 Pro or Wuji Glove. See the specification table for differences between models.

Combine with the LITCHI EGO egocentric data collection device

A data collection device that records hands-on work from a first-person view. Four wide-angle RGB cameras and one structured-light stereo depth camera capture tasks within 1.5 m in 3D, recorded in sync with T1 glove data.

RGB × 4 + depth camera × 1
Built-in 9-axis IMU and four microphones
Weight under 400 g

First-person video and depthA hybrid of four wide-angle RGB surround cameras and one structured-light stereo depth camera, accurately recording close-range work within 1.5 m.

RGB up to 1920 × 1200 @ 30 fps, global shutter
Depth 1280 × 800 @ 30 fps, spatial accuracy < 0.5% (at 1 m)
Weight under 400 g

Glove, video and audio in syncEquipped with a 9-axis IMU (500 Hz) and four microphones. Image-to-IMU sync error is under 1 ms, and external trigger signals are supported.

Image-to-IMU sync error < 1 ms
External trigger supported
Video encoding H.264 / H.265

Training-ready data from a single task

VisionFirst- and third-person RGB, raw and optimized depth, IMU.

Motion & contactHand joints, 3D pose and touch.

Space & objectsCamera pose, 6-DoF pose of manipulated objects, scene point cloud.

Task annotationTask instructions, subtasks, atomic actions, states and outcomes.

Use cases

Data collection for imitation learning of humanoids and dual-arm robots
Motion recording at real work sites such as factories, logistics and food processing
Teleoperation of multi-finger robot hands
Building first-person data for foundation models such as VLA

Product page ↗Request materials