Análise da Variação de Operadores de Mutação em Algoritmos Diferenciais Evolutivos para o Planejamento de Caminho de Robôs Futebolistas

Abstract

Path Planning for Mobile Robots is a relevant topic in autonomous robotics, being addressed in various areas, such as robot soccer. This work analyzes the performance of Evolutionary Algorithms, specifically Differential Evolution algorithms, in path planning, with a focus on the influence of different mutation operators. The algorithms DE, SaDE, JADE, SHADE, and L-SHADE were evaluated, considering the following mutation operators: "rand/1", "rand/2", "best/1", "best/2", "either-or", "current-to-rand/1", "current-to-best/1", and "current-to-pbest/1". Initially, the comparison strategy was tested on benchmark functions from CEC2013 and subsequently applied to path planning using NURBS curves. The experiments indicate that the L-SHADE algorithm, combined with the "current-to-rand/1" and "rand/2" operators, showed consistent performance in generating smooth and feasible paths for the soccer robots. The results suggest that the choice of mutation operator can influence the efficiency of certain algorithms in optimizing paths.