REDES NEURAIS PROFUNDAS E ALGORITMOS GENÉTICOS: ESTUDO DE CASO: TREINAMENTO DE CARROS AUTONOMOS

Abstract

Traffic in urban areas is one of the challenges of contemporary society. Increase constant number of vehicles, creates an environment conducive to accidents and congestion. In this scenario, the development of autonomous cars appears as a potentially revolutionary solution to improve road safety and optimize urban mobility. This work describes an approach to enable autonomous cars to travel in environments with obstacles, combining Deep Neural Networks and Genetic Algorithms. The goal is to develop an autonomous driving system that can learn to take decisions based on sensory information from the environment. A virtual environment was created to train autonomous cars using simulation scenarios with obstacles. The application of genetic algorithms optimized the process of training, allowing the neural network to adapt to different obstacle conditions. The results of the experiments demonstrated remarkable progress in autonomous driving. The integration of neural networks and genetic algorithms has enabled the self-driving car navigate with greater safety, anticipation and agility, minimizing the risk of collisions and optimizing driving efficiency. The approach results in steering systems autonomous vehicles more efficient and safe, contributing to the evolution of the field of mobility and road safety.