Implementación de un sistema prototipo de luces frontales con segmentación para automotores empleando técnicas de visión artificial difusas.

Abstract

Due to the large number of traffic accidents that occur at night due to recklessness and inexperience of drivers who do not perform the lighting raises from high to low, a diffuse artificial vision system was implemented for the respective segmentation of long lights, with the purpose of achieving non-glare towards other drivers who travel on the road The prototype construction was made with the conformation of a Raspberry Pi 3, Arduino ONE, Servomotors, LED bar, automotive relay and protection elements. For the development of the detection system, emphasis was placed on the detection algorithm "Haar Cascade", the graphic interface was developed in the Qt Creator environment. When the system detects the presence of a vehicle traveling in the opposite direction or in the same direction, it sends the angle determined by the rules of fuzzy logic to the Arduino causing the servomotors to move, achieving the respective light segmentation. The cascade classifier allows you to create a series of data admissible to the vehicle recognition determined the position of it. The graphical interface will be visualized by the user allowing to manipulate 6 buttons, each one fulfills specific functions such as: start camera, stop, detect, graph fuzzy, connect and disconnect. According to the tests carried out, it is established that the maximum distance at which a vehicle can detect is approximately 200 meters; in the same way, the system presents better results at low speeds due to a delay in the acquisition of the images in real time.