With the continuous development of artificial intelligence and robotics technology, stereo depth cameras, as crucial sensors, are being widely applied in various robotic systems. As indispensable perception devices for robots, their high-precision depth information provides robots with capabilities such as environmental perception, stereo vision, and pose recognition, enabling robots to achieve tasks like intelligent navigation, human-robot interaction, and object tracking.

Stereo Vision Camera

Recently, leveraging its technical accumulation and product experience in high-quality AI imaging, the company has launched a series of stereo vision cameras applicable to robots. This product consists of two IR image sensors, one RGB image sensor, one infrared speckle projector, and one IMU, and is a 3D stereo camera supported by a USB3.0 interface.

The stereo vision camera adopts the active stereo vision principle, enabling it to acquire depth information more accurately. This series of cameras not only features high-resolution, high-frame-rate image data but also boasts excellent imaging performance, providing high-quality images in various application scenarios, ensuring AI robots can see clearly in diverse environments.

The stereo vision camera supports both active stereo technology for 3D imaging based on structured light and passive stereo technology for 3D imaging based on visible light, enabling depth measurement up to 25m. This helps robots achieve stable and reliable perception capabilities in complex and dynamic environments.

The stereo vision camera can output image data up to 1280x800@30fps. Both the IR and RGB image sensors utilize global shutter technology, which minimizes motion artifacts, allowing robots to maintain high-quality perception performance even in high-speed motion environments. Additionally, this stereo camera integrates with an internal 6-axis IMU, providing high-precision 3D positioning data for robots in various application scenarios.

▲Stereo Camera Image Quality

The stereo camera combines stereo vision with an RGB camera. The RGB image data provides more basis and support for robot perception and decision-making. Robots can utilize RGB data to acquire characteristic information such as object color and shape, thereby achieving more precise recognition and grasping, and more comprehensively meeting the demand for highly flexible grasping operations in robots.

In terms of industrial design, the stereo vision camera features a miniaturized design. Its compact size of 90.75x25x30mm gives it higher compatibility, allowing for seamless integration into more robots and spatial perception applications. Furthermore, this stereo vision camera is widely applicable for AGV/AMR/drone obstacle avoidance, navigation, and other applications.

NVIDIA-based Stereo Camera Solution

As an NVIDIA partner, leveraging software driver technology on Jetson AGX Orin, we have deeply integrated and successfully adapted the series of stereo vision cameras with the Jetson AGX Orin platform. This will enable more developers to more efficiently develop their own applications, as well as conduct learning and research.

The stereo vision camera can connect to the Jetson AGX Orin platform via a USB3.0 interface and operate on the ROS system. Based on the Jetson AGX Orin development platform, a development kit for the stereo vision camera will also be launched later, providing full-stack technical support and services for developers in the field of robotic vision, for their high-precision 3D stereo camera development solutions, and supporting the deployment of robots in a wider range of applications.