Diseño e implementación de un prototipo de conducción asistido para conductores de automóviles en caso de deslumbramiento empleando técnicas de procesamiento digital de imágenes.

Abstract

This research describes the design and implementation of the assisted driving prototype using digital image processing techniques to help car drivers when they are dazzled by glare produced by another vehicle circulating in the opposite direction. Existing systems were initially investigated, and the effects of glare on drivers at night determined the prototype's requirements. An artificial vision algorithm was developed on the Python platform using tools with the OpenCV library for lane recognition and detect the light levels that can produce glare, an Alert signal protocol managed by a microcontroller on the Arduino UNO platform. Through positioning tests for equipment with image acquisition, it was verified 100% lane recognition efficiency when the camera is positioned on the side front of the car, 97.74% on the roof, and 90.09% between the hood and the windshield; the evaluation was carried out on an ideally marked road with straight paths, open and closed curves, generating margins of error were in cases that did not reach total efficiency. A Myo bracelet was used to acquire electromyographic signals located on the forearm of the driver to evaluate driver response with a visual stimulus, auditory, and an electric discharge, determining in natural conduction efficiency of 100% of the three types of alert signals and a simulated environment that, out of a total of 10 alerts generated randomly for each kind of stimulus, 100% of the auditory and electric shock were perceived by the driver, while the visual by 80%. In conclusion, the system of alerts combining the three experimental types would be the ideal one.