Transformación digital de una celda de manufactura mediante la integración de un brazo robot Industrial y visión artificial

Abstract

Abstract: This work addresses the digital transformation of a manufacturing cell through the integration of a xArm6 industrial robotic arm from UFactory and a computer vision system based on the Intel RealSense D435 camera. The objective was to automate the detection, classification, and manipulation of objects in an experimental laboratory environment in order to improve operational efficiency and system repeatability. The proposed architecture is based on a cyber-physical system composed of three subsystems: perception, robotics, and processing and control. At the software level, ROS 2 Humble, MoveIt, Gazebo, OpenCV, and Python were integrated for image acquisition, coordinate estimation, and robotic trajectory planning. The methodology included a simulation phase followed by an experimental validation phase conducted at the RobLab laboratory of the Universidad Católica de Cuenca. The results showed a success rate of 92% in simulation and 100% in the real environment, with average cycle times of 5.30 s and 5.60 s and positioning errors of 2.1 mm and 2.8 mm, respectively. These results demonstrate that the integration of industrial robotics and computer vision enables efficient automation of laboratory-scale manufacturing cells.