Desarrollo de una plataforma móvil tipo Ackermann para la validación de algoritmos de navegación autónoma dentro del campus ESPOCH

Abstract

The objective of the present study was to develop an Ackerman-type mobile platform equipped with the artificial vision to validate autonomous navigation algorithms within the ESPOCH campus. The live information was obtained through a pi camera, which sends information to a Raspberry pi, in charge of processing information and sending it to the actuators. The platform was developed in conjunction with a design and construction of the same; the validation of the algorithms was carried out in different environments and meteorological situations; covering stages of artificial vision for the detection of traffic signs and road detection. The first stage dealt with the detection algorithm through a pi camera developed in OpenCV. The second stage was responsible for the location of traffic signs and road detection through filters such as Canny, Haar Cascade, and the Hough Transform, resulting in an accuracy of over 85% in all tests under favorable conditions. For the statistical analysis, a data collection model of between 20 to 30 data in 3 stages of the day was used for a subsequent analysis, in Minitab, giving a higher percentage of effectiveness in the noon meals. It is concluded that the validation of the lane detection algorithms, traffic signals and traffic lights act correctly in a noon environment where the light favors the camera. It is recommended for a better reading and speed to adapt a more powerful camera such as a Raspberry with higher capacity and speed to avoid delays in live readings.