Cost Effective Machine Vision System for RealTime Defect Detection and Energy Efficiency in Learning Factories

Machine vision (MV) is a useful tool for industrial settings that aim for real-time data processing. Any industry may use this to monitor workpieces for geometric faults and material abnormalities in real time. Defect detection, workpiece sorting according to physical qualities, and process abnormality analysis are all made possible by using the convolutional neural network (CNN) method on real-time information from a low-cost, basic Raspberry Pi with a camera and the open-source ML platform TensorFlow. Using the planned cyber-physical production method, the cyber world may be connected to the physical units of a Learning factory (LF) to construct a data collection system based on MV. LFs are now often used in various sectors to train workers to handle the new technologies and problems that have emerged as a consequence of technological advancement. This study shows how to use an affordable MV device in a learning manufacturing setting to get data in real-time and save energy. It was discovered that the suggested inexpensive machine vision system can identify surface flaws, colours, and geometric imperfections. The straightforward and inexpensive MV system has improved energy efficiency and decreased overall carbon footprint by $19.36 \%$ and $798.22 \%$, respectively, depending on the MV method along the flow. Incorporating action-based learning methodologies into the classroom also improves the quality of instruction. This may reduce the need for humans and operators while assuring less rework, enhanced control, unbiased decisions, and complete quality assurance.