Uma heurística simplificada para funções custo de planejadores da família A*

• Ano de defesa: 2015
• Autor(a) principal: Silva, Jefferson Barbosa Belo da
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal da Paraíba; Brasil; Informática; Programa de Pós-Graduação em Informática; UFPB
• Palavras-chave em Português: Função custo; Planejamento de caminho; Geração de trajetória.; Cost function; Path planning; Trajectory generation.; CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: https://repositorio.ufpb.br/jspui/handle/tede/7846

One of the main issues related to the mobile robotics area is to find the most efficient way to perform the navigation from one point to another over environments, considering maximum safety and spending as less as possible time and computer resources. From this perspective, the aim of this work was to specify improved heuristics that could be applicable to cost functions of key A* based algorithms and use, more efficiently, the available computational resources. In this way, our approach aimed at minimizing the amount of collisions, the length of paths, and the processing time by minimizing the importance of g(n) term, which accounts for storing information from past steps of A* family algorithms. To show the effects of this modification, a survey of the best search strategies in dynamic and static environments was carried out and, after that, we analyzed the four best and latest algorithms, according to the specialized literature. Some comparisons have been made considering static and highly dynamic environments with different directions and search parameters to measure the quality of generated paths. Then, these algorithms were again analyzed with their cost functions modified according to our approach. The results of the comparison show that the R* algorithm, with forward search, is the most efficient for different spaces and searches. However, the change in their respective cost functions provided a significant improvement in the already excellent results achieved by the algorithms. In static environments, this modification showed up to be more effective for large and complex problems, which are commonly used for real robots. In highly dynamic environments, the cost function modification provided a considerable reduction in the time of planning and number of iterations to find the goal, as well as reductions in the memory utilization.