Research on the loading system of large-caliber micronano optical structure lithography machine

In the context of the industrial scenario for loading and unloading in large-caliber micro-nano optical structure lithography machines, the paper addresses the calibration of SCARA robotic arm's hand-eye coordination and the problem of target object detection and positioning. A method for SCARA robotic arm hand-eye calibration is proposed, which combines a full two-dimensional affine transformation with a gradient-based Hough circle detection algorithm. For target object grasping and placement, a solution based on YOLOv5 for target detection and positioning, along with dynamic path planning algorithm, is introduced. In this approach, the SCARA robot first captures image information through a visual sensor and feeds the image into the YOLOv5 deep neural network model, which identifies the object category and pixel coordinates in the scene. Combining this with the hand-eye calibration results, the pixel coordinates of the objects are transformed into spatial point information that the robot can execute. Finally, dynamic path planning is performed based on the confidence level of the detection results. Experiments demonstrate that the hand-eye calibration method, target detection and positioning, and dynamic path planning proposed in this paper can effectively carry out loading and unloading tasks for large-caliber micro-nano optical structure lithography machines, thereby improving system efficiency.