A.I. Deployment in Verification Tool for Inspection Equipment

Tan Piet Gek, Khoo Yak Hua, Ooi Jiwei, Chia Ming Wei, Wong Chia Ming
Vitrox Corporation
Penang, Malaysia

AI, or Artificial Intelligence, is a complex topic that is a rapidly growing field nowadays. At its core, a “true” AI is a machine that can simulate human intelligence, behaviors, and even emotions. As the use cases for artificial intelligence grow, it’s inevitable that we’ll discover more ways it can improve our lives. It has sprawled into various applications in industries such as manufacturing, finance, healthcare, e-commerce, social media, etc. We are living in the Fourth Industrial Revolution (a.k.a Industry 4.0), marking the rapid transformation of all manufacturing processes. The implementation of AI-based technologies has inevitably made huge changes in manufacturing.

The commonly known challenges in SMT manufacturing are achieving good yield, lowest operation cost, optimized throughput, etc. One of the main points here is that most processes still require human operation and a certain level of technical competency. Yet, it is high cost, slow, inconsistent, and uncertain performances. In this paper, we will discuss AI applications for Smart Manufacturing, focusing on the usage of inspection solutions such as AOI.

Generally, operators are required to perform results re-verification after AOI inspection. AI has been deployed in the verification tool with studies conducted to compare the effectiveness and efficiency of the tool versus human operation. The scope of this paper also includes classic AI modeling and different AI modeling deployments to handle different types of defects in SMT.

In the SMT industry, a significant problem arises from the inherent inconsistency in human performance due to emotional factors, behavior variations, and fatigue. Human operators, despite their expertise, are susceptible to emotional fluctuations, distraction, and personal factors that can impact their ability to maintain consistent levels of precision and focus during the process.

These inconsistencies may result in placement errors, and result in a lower overall product quality. Moreover, prolonged exposure to repetitive tasks can lead to operator fatigue, further exacerbating the issue. As a consequence, this inconsistency in human performance poses a substantial challenge in ensuring the reliable and high-quality production of electronic components, necessitating solutions that mitigate the impact of human variability on SMT processes. In the midst of Industry 4.0, where manufacturing processes are rapidly changing by adapting the AI technology to drive the changes, addressing the increasing demand for precision, efficiency, and quality.