Machine vision systems enhance the capabilities of industrial robots, enabling them to identify parts, perform quality inspections, read barcodes, and adapt to environmental variations. However, successful integration of these systems requires careful consideration, as the camera, lighting, robot controller, and PLC must communicate effectively and in real-time.

Choosing the Right Camera

Different types of cameras are suited for various machine vision applications:

  • 2D Area Scanning: Commonly used for data collection, inspections, and barcode reading. The CA and SC series from HikRobot are widely utilized.
  • 3D Structured Light: Best for tasks like box picking and measuring complex geometries, though it operates more slowly and independently of position.
  • Line Scanning: Effective for continuous inspection of surfaces such as tire treads and steel sheets, requiring movement along a conveyor.
  • Smart Cameras: The SC6000 series combines a camera and vision processor into one unit, removing the need for a separate computer.

Lighting: A Critical Factor

Lighting plays a crucial role in machine vision, often more so than camera resolution. A clear image from a budget camera is preferable to a poor-quality image from a high-end model. Important lighting principles include:

  • Utilizing directional lighting to enhance contrast (e.g., backlighting for silhouettes, ring lighting for surface defects, and diffuse lighting for reflective surfaces).
  • Maintaining control over lighting conditions to reduce environmental variability.

Communication Between Robot and Vision System

Effective communication between the robot and the vision system is essential. For example, in systems using FANUC and HikRobot, the integration typically involves the robot sending a digital trigger pulse to the camera, waiting for a "result ready" signal, and executing a Karel program to read XY coordinates from the TCP socket. This program writes offset values to numerical registers, which the TP program then applies to the collection position.

Calibration: Aligning Coordinate Systems

Before a robot can effectively use vision positions, it must calibrate the camera's coordinate system to match that of the robot. For fixed cameras (eye-on-hand), the robot should touch at least nine calibration points on a grid, recording the pixel coordinates for each. The resulting calibration matrix converts pixel XY coordinates to robot XY coordinates.

Xpert Robotics has integrated HikRobot vision systems with FANUC and Shibaura robots for applications such as quality inspection and box picking. For those interested in exploring machine vision integration, inquiries are welcome.