LivSyn · Handheld data capture
LivUMI
“Human-centric” capture accelerates embodied-model evolution
LivSyn handheld data acquisition for embodied intelligence
Beijing LivSyn Technology Co., Ltd.
A leading provider of embodied intelligence data infrastructure
Built on high-precision, ultra-lightweight mechanical design, millimeter-grade spatial pose tracking, sub-millisecond multi-sensor synchronization, one-click co-deployed software and hardware — LivUMI delivers high-quality embodied data to help world models generalize through real-world interaction and advance humanoid intelligence.
Use cases
From research and education to development and deployment — keep high-quality data flowing into your model iteration loop
Embodied data capture sites
Replace traditional mocap plus teleop rigs — lower cost, higher throughput for real-scene data
University & vocational labs
Go beyond 2D simulators — students run full 3D robot data workflows hands-on
Skill development platforms
Plug-and-play alternative to homogeneous teleop — faster access to high-quality data
Collect-and-train in the field
Industrial, service, and home-like settings — capture on site and train continuously at low cost
Why LivUMI
Easier
Ready out of the box, auto-starting stack, pre-screening and multi-format export, mobile status and recording at a glance
More portable
Ergonomic design, no robot body required — capture in diverse environments
More flexible
Modular gripper design — works with multiple arm brands and end effectors
Mechanical design
Lightweight aluminum transmission parts for precision and durability
Multi-sensor fusion
1 ms hardware-synced sensing
Vision
Ultra-wide fisheye and high-accuracy stereo depth
Gripper
Dense tactile array; 0.1 mm closure sensing accuracy
Specifications
| Parameter | Value |
|---|---|
| Dimensions | 160 mm × 200 mm × 180 mm |
| Weight | 460 g |
| End-effector sensing accuracy | 0.1 mm |
| End-effector jaw opening | 0–100 mm |
| End-effector grip payload | 2 kg |
| Tactile force resolution | 0.1 N (optional tactile fingertips) |
| Tactile spatial resolution | 1 mm |
| Depth camera | Depth FOV: 84° × 58°, ±2%; accuracy error: <2% (within 50 cm) Resolution: 1280 × 720 @ 30 fps |
| Wide-angle camera | FOV > 200°, 90 FPS |
| IMU | 9-axis, 200 Hz |
| Spatial positioning accuracy | 3 mm |