A ROBOTIC VISION SYSTEM WITH INCORPORATION OF ARTIFICIAL INTELLIGENCE FOR INDUSTRIAL APPLICATION

Abstract

Automation is required for domestic high volume manufacturing in today's competitive electronic industry. It will bring about high productivity and improve the product quality. In this respect, vision is becoming a key to the implementation of factory automation. The objective of this research project is to focus on the design and development of a low cost and reliable robotic visual inspection system for an industrial environment. Performance speed of the system will be the key consideration in the design of the hardware system. The design of the software algorithms will incorporate artificial intelligence (AI) techniques to improve the overall flexibility and reliability of the automatic visual inspection system. As a specific example, a new robotic visual inspection system for the detection of defects on the markings and leads of integrated circuit (IC) is developed. The hardware module of the inspection system consists of a mechanical feeder system, cameras, vision processors and fiber optic lightings. Timing control of the various sub-systems of the hardware module and unit alignment are the critical considerations in the design of the hardware module. The software module in the inspection system consists of two main steps. The first step consists of proper segmentation and labeling of individual regions in the image for subsequent ease in interpretation. Once every region in the image has been identified, the second step proceeds by testing different regions to ensure that they meet the design requirements. Features are then extracted from the previously processed image for verification- purposes. The automatic visual inspection system for IC is later modified and extended to the inspection of die-related defects on the surface of the wafer. As a specific example, the inspection of the wafer of Gallium Arsenide Phosphide die (GaAsP) is addressed. In order to overcome the problem of the speed of inspection, simple and reliable algorithms are extensively used to process the image data unceasingly while it is being acquired. To further improve the flexibility and reliability of this automatic inspection system, AI technique is incorporated in the system. A rule-based approach is adopted and incorporated into the current inspection algorithms. The rule-based method considers a pattern to be defective if it does not conform with the rules specified in the design. Thus, there is no need for direct comparison between the pattern under test and a reference pattern on a point to point basis. This technique has greatly improved the reliability of the inspection process. This type of technique is also very flexible as it can easily be modified to inspect any complicated non-overlapping pactern. In summary, the robotic visual inspection system developed in this research project has been tested to work satisfactorily in a simulated on-line manufacturing environment. The results obtained are included and discussed in detail in this thesis.