A vision and robot based on-line inspection monitoring system for electronic manufacturing

This paper presents a neural network-based vision inspection system interfaced with a robot to detect and report IC lead defects on-line. The vision system consists of custom software that contains a neural network database for each of the ICs to be inspected on a PCB. The vision system uses gray scale images and a single layer neural network with three outputs based on defect criteria. Each IC has a different inspection area, thus, the input vector varies for each of the ICs. The IC networks were trained with Matlab's Bayesian regularization module. Performance of each of the networks investigated was found to be 100% based on the defect criteria. This system has been implemented and tested on several electronic products using ProE, C++ and OpenGL software platforms [R. Balderas, S. Bose, Automated robotic inspection system for electronic manufacturing, MSE Thesis, Manufacturing Engineering Department, UT-Pan American, 2002; A.I. Edinbarough, J. Amieva, Experimental study on the robotics vision inspection of electronic components, BS Thesis, Engineering Technology Department, UT-Brownsville, 2002].