Implementación de un prototipo de reconocimiento automático de líneas urbanas y geolocalización de la ciudad de Riobamba para personas con discapacidad visual utilizando visión artificial.

Abstract

In the this research was implemented a prototype of automatic recognition and geolocation of urban lines in Riobamba city for people with visual disability using digital vision. This prototype consists of four stages. In the first stage was used a GPS receiver, which provides longitudelatitude, compares with the Coordinates stops previously stored in a micro SD module and through the voice assistant sends an alert of a possible stop. In the second stage, the PixyCam 2 was used to detect the bus lines; three types of identification were performed: detection of the bus's own labels, detection with an identification placed on the front part of the bus and also the detection of the bus numbers through YOLO. In the third stage, a GSM SIM 900 module was used to send the coordinates of the blind person. In stage four, two Arduinos connected in Master-Slave mode were used to play mp3 tracks previously recorded from a voice narration application. In the positioning tests it was concluded that the range of difference between the GPS reading and the coordinates of the stored stops are from -5 to 5 meters. In the bus line detection tests, it was determined that detection process two is the one that provides accuracy and fluidity needed for the prototype, due to the color identifiers placed on the front part of the bus. The maximum detection distance between the camera and the bus is 8 meters. It is recommended to incorporate a camera with digital vision with better image reading in order to obtain better results in identifying the signs on the bus line.