Automated Box Classification in a Virtual Industrial Environment Using Machine Vision Algorithms

This article presents the development and evaluation of an automated system for box classification based on the detection of some ISO 780:2015 symbols, using a UR5 robotic manipulator and advanced machine vision techniques. The methodology integrates sensors, vision algorithms, and robotics to achieve precise classification. The UR5 manipulator, with six degrees of freedom, was kinematically modeled using Denavit-Hartenberg parameters, allowing for the calculation of the end-effector’s position and orientation. The system employs the Speeded-Up Robust Features (SURF) algorithm for the detection and classification of symbols on the boxes. Simulation results in CoppeliaSim, processed in MATLAB, reveal an average detection time of 55 milliseconds and a recognition rate exceeding 95%, even under varying lighting and orientation conditions. The system demonstrates high precision and flexibility in handling boxes of different sizes and arrangements, validating its efficiency in automated virtual industrial environments.