Levantamento e avaliação de funções score para algoritmo minimax em batalhas Pokémon

Abstract

The gaming market has gained more and more prominence in the world, with Pokémon being one of its main franchises. This way, the design of strategies for decision making in Artificial Intelligence agents is a constant concern of developers. So, this work aims to evaluate score functions for the Minimax algorithm applied to pokémon battles, in order to verify which approaches obtain better results. Using tools such as Pokemon Showdown and Poke-env, several simulations were performed to test the performance of the proposals presented in the literature. At the end of the experiments, despite the low number of proposals, it was possible to verify that evolution functions which were based on the percentage difference of Life Points between players obtained better results in the experiments, as well as those that contained a scoring system for status and effects.