Skip to Content

Build Advanced Robotics with Embedded Stereo Vision


Stereovis offers a fully integrated development experience, from edge devices with NVIDIA Jetson to SDKs, APIs, and ready-to-run examples in Python, C++, and ROS. Everything you need is right here.

⚙️ Quick Start


✅ Supported Languages

  • Python 3.7+
  • C++
  • C#
  • ROS (Noetic / Humble)

✅ Supported OS

  • Ubuntu 20.04 / 22.04
  • Windows 10/11 (x64)

🔌 Device Setup

  1. Connect the camera via Gigabit Ethernet
  2. Set your PC’s IP address to match the subnet
    e.g.: Camera 192.168.1.251, PC 192.168.1.254
  3. Enable jumbo frames + full duplex (recommended)

🔧 SDK Installation

Python:

 pip install stereovis_sdk

C++: Link the included .lib or .so from the /lib folder

Include headers from /include

C#: Add reference to StereovisAPI.dll in your project

📚 SDK Structure

/lib          → Precompiled libraries (Windows/Linux)
/include      → C++ and C# headers
/demo         → Sample applications
/tools        → Image viewer + config tool
/docs         → Full documentation PDFs

🐍 Python Sample

 

from stereovis_sdk import Camera

cam = Camera(ip='192.168.1.251')
cam.initialize()
cam.trigger()

pointcloud = cam.get_point_cloud('6D')
print("Points captured:", len(pointcloud))


💻 C++ Sample  


init3DCam(0, "192.168.1.251", imgW, imgH);
softTrigCam(true);
getXYZInt(150, 120, outXYZ);

🤖 Robot Communication  

  Send Pose via TCP (Python):

cam.send_robot_pose(x=300, y=0, z=200, roll=0, pitch=0, yaw=90)

Works with:

Universal Robots, ABB, KUKA, Doosan, Kinova, and many more mainstream 6-axis robots 

🔬 3D Output Options


  • getPtCloud4D() → XYZ + Intensity
  • getPtCloud6N() → XYZ + Normals
  • getPtCloud6C() → XYZ + RGB
  • getXYZInt(x, y) → 3D from pixel
  • getRect3DCloud() → ROI point cloud

📦 ROS Integration

Install via package manager:  

sudo apt install ros-noetic-stereovis

Launch:

roslaunch stereovis camera.launch

🧠 AI & Inference (Jetson Devices)

Use torch, onnxruntime, or TensorRT for inference:

from my_model import Net
model = Net()
output = model(cam.get_point_cloud())

Build boldly. Automate precisely. Create with vision.