Implementación de una estación para clasificación y organización de piezas plásticas basada en un robot cartesiano y visión artificial

Abstract

The objective of this research was to implement a workstation for the classification and organization of plastic parts based on a Cartesian robot and artificial vision. With the help of a technological methodology starting with the structural design and stress testing of the Cartesian robot with SolidWorks software, where the design of the elements was approved with a safety coefficient greater than 1. Followed by the selection of the hardware elements as a development board, a Raspeberry Pi 3 Model B, Arduino Mega, Nema 17 stepper motors, two types of controllers, a linear actuator for the axial axis of the robot and sensors were used. A machine vision algorithm was designed in Phyton for shape and color detection and identification of empty and full spaces, where libraries were used to digitize the image, thresholding, filters and matrix operations. The graphic interface serves as a machine-user communication where it is visualized in real time, also generating a database that allows the registration of the packs that have been filled. Where it was determined that the station presents a 100% result both in the recognition of spaces in the packs and the detection of shape and color of parts with a certain number of lux and it is verified with a percentage error close to 0 that there are no delays in the mobility between locations. Therefore, it is concluded that the implemented workstation is suitable for the classification and organization of plastic parts with which artificial vision and storage tests could be employed. The implementation of the workstation in warehousing logistics processes as a mechanical means is recommended.