Desarrollo de un prototipo electrónico que guíe a persona invidentes en la recolección de objetos mediante audio y visión artificial

Abstract

In this document we present the work done to develop a prototype that guides visually impaired people through audio and artificial vision. For this purpose, a system was implemented based on a Raspberry Pi 3 model B + electronic card that acquires images of two webcams connected to its USB inputs and processes the captures to recognize a green object by marking its center with a red box. The given center is used in Cartesian coordinates (x, y) to calculate the angle of deviation of the object with respect to the web cameras. In addition, the stereo vision of the prototype is used to calculate the distance of the object with respect to the two video cameras. For this last case, experimental measurements were made to determine the equation that will calculate the distance. With the help of audio signals, people are guided to correct the angle of deviation of the object with respect to the cameras, and with the calculated distance, voice messages are generated that inform the user about the distance to the object. In addition, the results are presented in the tests carried out with the prototype, which were 70% assertive so that the person manages to take an object, using an average time of 4.8 seconds among the users and with a relative error of 3.7%. Through these satisfactory results, the authors' conclusions and recommendations regarding the operation of the prototype are presented.