;Abstract Automation occupies a pivotal role today, especially in the manufacturing industry, each manufacturer has its own automation-related processes and equipment. This research is to try to implement automation on the R&D side, so the complex test items can be automated using scripts or graphic recognition systems to achieve the goal of improving test efficiency and quality. On the research and development side of the server industry, it is divided into several different verification stages. The foundry manufacturers must perform the same and highly repeatable tests at each stage. Therefore, if automated tests can be introduced, human resources can be reduced appropriately. In this study, we analyzed the feasibility of automation through case analysis, and using various tools to try, finally find a suitable method, mainly script control and supplemented by graphic recognition via Sikuli, and define the standard process, and then do the system modularized for automated imports in this general direction. Afterwards, the improvement of efficiency is calculated through the statistics of actual import and test time. This research is based on the actual execution and statistical results of the R&D end test of the case company to reach the following conclusions: 1. After the automation script matures, it can save about 50% of the time. 2. The consistency of test data is improved. 3. Reduce the error of human subjective judgment, so that the test quality is relatively improved.
