Libra1000&3000Series

AI stereo depth cameras

Product and sample inquiriesDatasheets and resourcesView specifications

The Libra 1000 & 3000 series brings passive stereo 3D vision to embodied AI robots. Deep neural network stereo technology provides 3D perception from 0.1 to 10 m, improving reconstruction of low-texture, reflective and transparent objects. High-frame-rate, real-time operation on embedded platforms supports autonomous mobility and precision manipulation.

AI stereo vision for challenging objects

Deep neural networks improve 3D reconstruction of challenging objects, including low-texture, reflective and transparent surfaces.

High-quality depth. Rich spatial detail.

High-quality 3D depth data provides the spatial information robots need for obstacle avoidance, grasping and pose estimation.

High-speed data over GMSL 2.0

GMSL 2.0 high-speed output supplies visual data for SLAM, mapping and obstacle avoidance algorithms.

Applications

Embodied AI robotics application

Libra 1000 & 3000 Series

Swipe to view all models and specifications

ModelLibra 1000Libra 3000
Product imageLibra 1000 product imageLibra 3000 product image
Use casesManipulation perceptionSpatial perception
RGB FOVH91° × V78°H115° × V99°
RGB resolution @ frame rate1280 × 1080 @ 30 fps1280 × 1080 @ 30 fps
Depth resolution @ frame rate640 × 540 @ 30 fps640 × 540 @ 30 fps
Stereo baseline20 mm65 mm
IMU6-axis IMU6-axis IMU
InterfaceFakra GMSL 2.0Fakra GMSL 2.0/
USB 2.0 TypeC
ShutterGlobal shutterGlobal shutter
Depth performance<1 mm @ [4 cm–20 cm]
<1% @ (20 cm–1 m]
<1.5% @ (1 m–1.5 m]
<1% @ [10 cm–1 m];
<1.5% @ (1 m–2 m]
<2% @ (2 m–5 m];
<5% @ (5 m–10 m]
Depth processing platformOrin NX/AGXOnboard processing
Explore product support

Building Visual Perception Infrastructure for Physical AI

Contact Us