Automatic Shape-Based Sorting of Ceramics Using a Robotic Arm with a Vision System

This study applies robotics arm and vision system technology as a visual system for ceramic product inspection using modern cameras. The system reduces employee inspection errors and expedites the inspection process caused by the differences in the quality of the workpieces. There is a negligible difference between NG and OK workpieces, making visual inspection difficult. Additionally, the error rate increases when employees visually assess the products, increasing inspection expenses. Ceramic products are classified into five categories: Small, Medium, Large, Covered, and Wide. This study proposes a 5-Degree-of-Freedom (5-DOF) robotic arm with a machine vision system incorporated for an autonomous shape-based ceramic sorting system. For training and validation, a dataset of seventy ceramic objects, bowls, plates, and cups was utilized. Pick-and-place tasks are carried out by the robotic manipulator using the categorization output. The experimental results demonstrate that the system achieves 90% classification accuracy and a sorting speed of 10 pieces per min, indicating an 80% increase in efficiency over human sorting. The proposed approach enables the adoption of smart manufacturing in the ceramic industry and improves production reliability.