AI Vision

Cambrian Vision

Cambrian Vision was developed for cost-efficient, fast and robust robotic bin picking, pick and place, machine feeding, feature location, assembly, supermarket and warehouse logistics and other tasks that require accurate robot vision.

Benefits of robotic AI vision include:

• Suitable for a wide range of parts types including transparent, shiny metal, plastic
• Has been designed to be robust for use in different external light conditions. It works in sunlight, artificial light and “disco” lighting
• Prediction times of approximately 0.2 seconds
• Can pick and detect parts with sub mm levels of accuracy

How does Cambrian Vision work?

1. Robot program (Robot controller) asks Cambrian for predictions.
2. Cameras take a set of images.
3. Images are routed through the AI model.
4. The AI model returns ‘pickable’ parts with their 3D position (x,y,z, a,b,g).
5. Robot gripper is moved to the predicted pickpoint, with collision checking.
6. Robot program actuates the gripper and picks.

What's included?

• Computing unit and the pre-installed Cambrian software
  Powerful GPU computer runs the Cambrian software and AI models for the part detection and communicates with the robot.
• Camera module
  Cambrian camera module uses high quality off-the-shelf industrial cameras.
• Accessories
  Calibration board, network cables and switch, Cambrian Unimount (a mounting plate for common ecosystem grippers).

Key Features

• Prediction time: 200 ms
• Minimum part size: 1 mm x 1 mm xx 1 mm
• UR+ Certified

Applications:

• Pick and place
• Bin picking
• Object detection
• Cable handling
• Kitting
• Assembly