External visual observers for understanding manufacturing operations & predicting faults

点击翻译按钮获取中文摘要

ATS Automation Tooling Systems Inc. provide solutions for factory design and automation. Their life sciences division provides manufacturing solutions for world leading companies engaged in medical devices, pharmaceuticals & diagnostics. ATS also develops automation systems for the transportation, energy as wellsthe consumer & electronics sectors. The Vision Image Processing (VIP) lab at the University of Waterloo isdedicated to understanding visual processes and finding solutions for the outstanding problems in image andvideo processing and intelligent systems. ATS currently utilizes camera sensors in machine vision monitoringof existing processes in addition to a collection of other sensors.After a manufacturing machine is produced, it is difficult to retrofit an existing machine with new sensors. This project will explore external observing cameras that can be placed around the machine and learn what is the normal operation and be able to detect and predict anomalous operations. There are three machine scenarios of interest: (1) a conveyor system; (2) an assembly machine; and (3) a bowl feeder machine.however VIP hopes to improve robustness and accuracy by providing AI based algorithmsthat understand the process including the visual scene landscape of the manufacturing process components andparts.