Color Classification Using a 3-DOF Robotic Arm Based on the YOLOv5 Model

Abstract

The combination of robotic arms and computer vision has greatly enhanced the efficiency and accuracy of object classification tasks across multiple industries. This work proposes the implementation of a 3-degree-of-freedom robotic arm designed for classifying objects by color using the YOLOv5 deep-learning model. Traditionally, object classification has been addressed through manual sorting or basic automated systems with limited accuracy. Our approach employs YOLOv5s for real-time color detection and provides the flexibility of automatic operation as well as remote control via radiofrequency. The system builds on recent developments in robotics and computer vision to enhance sorting accuracy. During testing, the model achieved 87% precision, demonstrating effective object classification and manipulation by the robotic arm. These results suggest significant potential for improving automation in color-based sorting