Bridging the Gaps: A Flexible Automated Object Classification System for PLC, Machine Learning, and Industry 4.0 Education

Industry 4.0 manufacturing increasingly relies on integrated cyber-physical systems. However, lab experiences in many college classrooms often present vision and controller subsystems as closed appliances. This can limit the ability of students to observe, validate, and troubleshoot behavior from start to finish. This paper presents a laboratory module and instructional station deployed in the classroom that connects computer vision inference in real time to programmable logic controller (PLC) actuation through a deliberately transparent supervisory interface. The platform integrates a standard webcam, a VB.NET inference application using transfer learning, an Excel workbook enabled by macros for explicit mapping of labels to commands, and PLC ladder logic that drives discrete indicator outputs. The module was delivered over six sessions of two hours each as part of a course on manufacturing automation and robotics. Engineering Technology students worked in small teams, and most achieved the intended closed-loop behavior. Real variations in capture conditions—such as lighting, clutter, pose, and occlusion—provided an immediate basis for debugging across every layer of the system. The results indicate that the module is feasible for delivery within typical laboratory constraints. Furthermore, the module helps learners develop a system-level understanding of how perception is integrated with control, using interface observability as a central design principle.