Computer-Integrated Micro-Assembling with Image-Servo System using ANN

A micro-assembling task using an image-servo positioning system is studied in this work. To improve the micro-assembling efficiency using manual operations, an image-servo automatic assembling system is established using a XYQ stepping positioning stage with a self-developed image recognition system. The circle outlines and the cross marks of the micro parts are used for this proposed system to control the XYQ-stepping stage to achieve assembling tasks. For the proposed image-servo system, the coarse positioning task and the fine positioning task are designed. First, the Sobel operator is used to find circle outline of positioning mark for the coarse positioning. Second, calculating the centre and radius of the positioning circle using a least-mean-square error is to guide the XYQstepping stage to perform the positioning. After performing the coarse positioning task, Artificial Neural Network (ANN) systems are studied to improve the positioning precision via compensating positioning errors due to the image distortion. The main contributions of this paper are using BP and RBF neural networks to perform the nonlinear geometry transformation from image coordinates of the pixels to the actual positions in the global coordinate system.