Redes neurais profundas na computação de heurísticas para algoritmos de busca de caminhos em mapas virtuais contendo elevação e inclinação

Abstract

Deep neural networks and pathfinding algorithms have been investigated in the area of Artificial Intelligence (AI). Despite this, these areas of research still require greater integration, mainly with a view to proposing pathfinding algorithms that explore relief information in route computations with lower distances and topographic costs. This work uses techniques of deep neural networks in the construction of heuristic functions used in the optimization of pathfinding algorithms that explore elevation and inclination represented in large virtual maps. Experiments compared such heuristics with traditional heuristics in the execution of the A pathfinding algorithm. Results showed that the use of deep neural networks can reduce the computational cost of the pathfinding algorithm in virtual maps containing topographic information.