Design and development of an automated inspection system for vials

Abstract

Automated inspection systems are powerful tools that can significantly reduce risk and maintain product standard. At Waters Corporation, the existing packaging process of a certain vial product relies mainly on manual labor. Designing an automated packaging machine is believed to greatly improve both packaging speed and quality. To ensure that the packaging machine produces proper packages, an automated inspection system is designed as the last stage of the packaging machine. Inspection is based on computer vision and image processing enabled by Raspberry Pi and OpenCV. The proposed algorithm is designed to determine if the proper number of vials are present in each package. Moreover, a data-logging and data-transfer channel is developed between the machine and Waters Corporation. Some basic Industry 4.0 features are also loaded so that machine status and operation data can be access remotely in real time. This thesis presents an overview about the automated inspection system, details about the computer vision algorithm as well as the industry 4.0 feature. For other subsystems, please refer to other theses from the 2019 MIT Master of Engineering in Advanced Manufacturing and Design Cohort.