Flexible Automated Optical Inspection Architecture for Industry 4.0

The industry of the future, namely Industry 4.0, is built on top of novel and automated technologies, which optimize the production process by empowering industrial machines and process with novel and tailored software components. Artificial Intelligence algorithms enable to automate and improve traditional operations in industry, by learning from available data patterns and offering optimized operations. This scenario presents different research challenges: at first, there is the need for an architecture which can be customized depending on the specific task; at second, the knowledge of the system should grow together with the users, which need to instruct it according to their needs. Moreover, the real deployment of these system still faces challenges such as the latency and accuracy. We center our analysis on the Automated Optical Inspection task, and we present a novel architecture which faces such challenges. Our results show how these systems can be deployed in real scenarios and under which constraints they need to operate.