Development of a vision-based automated hole assembly system with quality inspection

Abstract

This study proposes a novel mechatronics system capable of automating a peg-in-hole assembly process and inspecting the quality of the assembly with vision. The proposed mechatronics system integrates a customized peg insertion tool, a new assembly mechanism, and a control algorithm to efficiently insert pegs into holes with a tolerance of 200 µm. The system improves assembly performance by utilizing dual cameras and several computer vision techniques, including Contrasted Limited Adaptive Histogram Equalization, Canny Edge Detector, Hough Circle Transform, and YOLOv5. In addition, a real-time statistical quality inspection method is proposed and compared with machine learning-based inspection approaches. Various surface textures and materials of the cylindrical workpiece are used to assess the robustness of the proposed inspection methods. Experimental results demonstrate that the proposed system and quality inspection methods exhibit high performance and adaptability.