INTELLIGENT ROBOTIC SYSTEM FOR AUTOMATIC IDENTIFICATION OF MARKS ON PARTS.

The article focuses on the development of an intelligent robotic system for the automatic identification of marks on parts, integrating a vision system with a Delta-type parallel robot to perform pick-and-place operations. Using a low-cost Logitech C270 webcam and a convolutional neural network (CNN) trained on the EMNIST ByClass dataset, the system identifies letters printed on cubes to form words, simulating industrial mark recognition. The CNN model, designed with eight layers and optimized through specific training strategies, achieved 92.87% accuracy on validation data and 100% accuracy in practical tests under varying lighting and placement conditions. This approach demonstrates that effective mark identification in industrial settings can be achieved without expensive cameras or complex algorithms, offering a cost-effective alternative for Industry 4.0 automation.