Hand-Eye Calibration Concepts

The method used by the robot to communicate with the vision system. The Mech-Mind vision system supports three communication methods: standard interface, Adapter, and master control. For details, refer to Communication Overview.

The way the camera is mounted in the workcell. Common mounting methods include Eye to Hand (ETH) and Eye in Hand (EIH).

Depending on whether the process of collecting calibration images and poses is automated, calibration methods are divided into automatic calibration and manual calibration. Manual calibration is relatively complex, so automatic calibration is recommended.

During calibration, connect the robot. Mech-Vision automatically plans the calibration path, controls the robot to move along the planned path, and collects calibration images and poses.

During calibration, the robot is not connected. You must manually control the robot to move along your planned path or touch calibration circles, and manually enter robot poses and trigger calibration image capture.

The method used to collect calibration data. Mech-Vision supports only the Multiple Random Poses of Calibration Board method.

The robot passes through path points that are automatically generated by the software or manually planned, and at each path point the system collects calibration board images, identifies calibration circles, and collects robot flange poses. This finally forms a set of calibration points and establishes the relationship among the calibration board, camera, and robot. This method is simple and provides high calibration accuracy. It is recommended for 6-axis or 4-axis robots.

Camera intrinsic parameters are basic internal camera parameters, including focal length and distortion. In general, intrinsic parameters are calibrated at the factory and stored inside the camera.

Camera extrinsic parameters describe the pose transformation relationship between the robot and the camera (that is, the hand-eye relationship; therefore, robot hand-eye calibration is also called camera extrinsic calibration). The relative pose between robot and camera is not fixed across different applications, so on-site calibration is required to obtain the hand-eye relationship.

A marker is an object used in place of the calibration board during hand-eye calibration. A marker should have clear feature points and evenly distributed feature points.

The robot pose when the camera captures a calibration board image during calibration. When calibration data is collected using the Multiple Random Poses of Calibration Board method, calibration points are the path points in the calibration path.

The circular feature point on the calibration board. During calibration, the software calculates the pixel coordinates of each calibration circle center and the circle center coordinates in the camera coordinate system from captured calibration board images (2D images and depth images), and then computes camera extrinsic parameters based on the collected calibration circle data.