Diseño de un sistema de control para clasificación de objetos utilizando un brazo robótico KUKA KR10 R900 SIXX mediante visión artificial.

Abstract

This research aimed to develop a control system for object classification using a KUKA KR10 R900Sixx robot with artificial vision. The system integrates two parts: the software allows for detecting objects based on their characteristic color (red or blue), and the hardware executes the necessary movements to transport the objects. In addition, the system integrates a template that houses the objects to be classified, located within the working area of the manipulator. The hardware and software are physically interconnected through a point-to-point local network between the computer and the robot controller, while at the software level, the connection is established through the Kukavarproxy controller within the RoboDK software, designed for KUKA-manufactured robots. This software is a communication interface between the system's software and hardware. In addition, it enables the integration of various algorithms for both vision and motion control using Python as the programming language. The tests for color detection to determine correct operation at various speed levels have allowed the system to be validated by verifying the existence of a direct relationship between the detection variables and system illumination through statistical analysis based on hypothesis evaluation. This has resulted in a 95% accuracy rate for color detection through confusion matrices as a statistical metric. This way, it has been determined that the system works effectively based on the objectives and requirements proposed, with execution times ranging between 46.32 and 47.38 seconds.