Desarrollo de un sistema de control para un brazo robot que clasifique objetos en base al color utilizando visión artificial.

Abstract

This document explains the development of a control system for a robot arm that classifies objects based on color using artificial vision. This document presents a solution to the problem of manipulating objects using a robot manipulator arm "xArm" from LewanSoul, which is controlled and guided based on the color recognition of an object using MATLAB computer vision libraries. To solve this problem, the trial and error research methodology was used through the spiral method that is based on solving a problem in four phases starting with planning, followed by risk analysis, implementation and ending with evaluation to observe the results achieved and start again with the spiral if the results are not as expected. In the first phase, the objectives were planned and delimited, which were to recognize the color of the object to be manipulated using artificial vision techniques with the help of MATLAB and a Logitech C930 camera. As a second phase, the mathematics of the robot was analyzed to guide it to the point that defines the location of the object to be collected, in the same way the strategies for the creation of the graphical interface and the communication between software and hardware were analyzed. In this analysis, the types of controls needed to command the arm were evaluated, and a MATLAB GUI was used to create the graphical interface, in addition, serial communication was chosen to send data to the LewanSoul servo controller card. In the third phase, the system was implemented connecting the LewanSoul arm with the MATLAB PC and the Logitech C930 webcam through serial communication. Finally, the results obtained were evaluated and it was found that the system presents 97% fidelity in the control of its joints from the graphic interface, 100% in the recognition of colors (red and green) and 97% in terms of to the classification of objects, with minimal errors due to the PC processor that were solved when executing the actions again from the first phase of the spiral method. In the end, the job is finished concluding that the robot is reliable for use in industry within repetitive sorting processes.