Hi,您当前浏览器版本过低，可能会影响浏览效果，建议下载使用感更好的浏览器，或者使用360浏览器极速模式浏览!
中 | EN
Shop
Space Camera is a native RGB-D spatial camera built for Physical AI. Powered by RoboSense’s proprietary Peacock SPAD-SoC and a fully solid-state architecture, it is ultra-compact and robustness, delivers pixel-level RGB-D alignment, synchronized IMU data, and millimeter-level depth precision. By natively outputting measurable, reusable multimodal data, Space Camera faithfully captures real-world color, geometry, and spatial position—providing high-quality data for embodied intelligence, spatial reconstruction, and model training.
More Capability
One Compact System
Powered by the proprietary ultra-large-array Peacock SPAD-SoC and an fully solid-state architecture, Space Camera packs pixel-level RGB-D alignment and millimeter-level precision into a compact design, turning multimodal data from merely usable into truly actionable.
  • Native Color Point Clouds, No Registration Required
    RGB images of up to 12 megapixels are pixel-level aligned with million-point 3D point clouds, reducing the need for extra registration and delivering high-quality data that scales easily for object recognition and semantic understanding.
  • Millimeter-Level Precision, Fine Details Preserved
    Direct depth output eliminates the need for computational depth estimation. Depth precision of 5 mm (1σ)*, preserves finer details and enables more precise localization and manipulation, enabling precise real-time perception and manipulation and helping reduce accumulated trajectory drift.
  • Flexible Outputs for Diverse Workflows
    Multiple output combinations — free combination of RGB, depth, and grayscale images, plus customizable high and low frame rates (Frame rates of up to 60 Hz) — flexibly adapt to perception, data collection, and algorithm development.
  • Wide FOV, Long Range
    A 120°×90° fused FOV and ranging of up to 50 m, combined with power consumption of ≤5 W, provide broader spatial coverage for large-scale capture and mobile perception.
  • Fully Solid-State, Built for Flexible Deployment
    With no mechanical scanning components, its fully solid-state architecture with high reliability. Measuring just 75.9 × 44.9 × 36.1 mm and weighing only 160 g, it integrates easily into head-mounted, handheld, and wrist-mounted data-capture systems.
  • Proprietary SPAD-SoC, Real-Time 3D Imaging
    With no mechanical scanning components, its fully solid-state architecture combines high reliability with a compact, lightweight design, enabling easy integration into head-mounted, handheld, and wrist-mounted data-capture systems.
High-Quality Data
Unlocking Better Algorithms
Space Camera delivers high-quality multimodal data for reconstruction, tracking, training, and localization, helping algorithms achieve more accurate and reliable results. Native RGB-D alignment also reduces dependence on complex data acquisition and post-processing registration, enabling developers to tackle more complex environments and more precise robotic tasks.
  • 3DGS
    Reduces reconstruction gaps, floating artifacts, and structural distortion in weak-texture, changing-light, and occluded regions, producing sharper geometry with consistent real-world scale.
  • 4DGS
    Synchronized multimodal data with low latency and consistent multi-view geometry reduces misalignment and ghosting around moving objects, enabling more faithful reconstruction of motion in 3D space.
  • Motion Capture
    Continuously captures body, hand, and finger-joint motion and interaction, providing accurate, dense data for pose estimation, fine manipulation, and more efficient training-data preprocessing.
  • RGB-D SLAM
    Reduces point-cloud layering, blurred edges, and structural misalignment, maintaining stable mapping and localization under rapid motion, shaking, abrupt viewpoint changes, weak textures, and dynamic scenes.
  • Perception and Obstacle Avoidance
    Millimeter-level depth information enables reliable detection of fine obstacles such as cables, step edges, and small scattered objects, improving autonomous obstacle avoidance in complex environments.
Built for the Real World
Advancing Physical AI
With high-quality multimodal data, Space Camera supports more accurate environmental understanding, finer action learning, and more realistic spatial reconstruction — driving embodied intelligence and spatial intelligence toward broader practical applications, and accelerating the deployment, iteration, and evolution of Physical AI across more scenarios.
AI-Ready, Boosting Development Efficiency
The AI-Ready ecosystem provides a complete set of developer resources, including documentation, multi-scenario datasets, and platform support. Compatible with NVIDIA, Rockchip, Horizon Robotics, and other computing platforms, it shortens development cycles and simplifies deployment—accelerating Physical AI from prototype validation to real-world applications.
Explore AI-Ready Specifications
*Based on empirical values from general-scenario testing, ranging accuracy may vary with distance and target reflectivity.
中 | EN
  • RoboSense Technology Co., Ltd (2498.HK)
  • service@robosense.cn
  • 0755-86325830
  • North America Offices
  • 46057 Five Mile Rd., Plymouth, MI 48170
  • 800 West El Camino Real, Suite 180, Mountain View, California 94040
  • rsi-office@robosense.ai
  • Whistleblower Email:voice@robosense.ai

Privacy Policy    Copyright © RoboSense

This website uses cookies ensure you get the
best experience on our website.Learn more.

I accept cookies I refuse cookies