Desarrollo de un sistema basado en visión artificial para manipulación sin medio físico de un modelo de planta dispensadora y mezcladora de líquidos.

Abstract

This research describes the development of an artificial vision based system for the non- physical handling of a liquid dispensing and mixing plant model. The starting point was the design of the prototype plant using the CAD tool Solid Works, plans were made and a static analysis of the parts under stress was carried out for their validation. LabVIEW was used to create a computer application that centralizes system resources. The vision Acquisition and the Vision Assistent packages of the LabVIEW Software allowed the encoding of a vision algorithm based on the correlation of images for the identification of gestures generated with a person´s hand. A section was generated for the creation of control signals directed through sequential communication to an Arduino that, together with a Shield CNC, manages the handling of the operators of the protitype plant. The implementation of the plant was achieved by determining the possible execution of seven sub-processes controlled by a specific gesture and numbers of credits, achieving simple dosages, double and triple mixing of low intensity liquids. The efficiency of the vision system in detecting gestures was determined through 50 continuous tests, achieving an average efficiency of 93.14% and an average response time for gesture recognition of 1.81 seconds. It was also proven that the prototype has a constant and homogeneous functionality, by timing in ten tests an average time of 22 seconds in the transport system to move the container trolley from one station to another, with an average time of 14 seconds for the dosing process. It is recommended to implement a lighting control system, to avoid continuous calibration of the artificial vision algorithm.