Raciocínio baseado em casos de simulação

Abstract

The use of simulation systems to train people is fundamental in different areas. Artificial Intelligence (AI) techniques can be exploited in the design and implementation of intelligent resources to provide more realistic and challenging behaviors to the different elements that make up these simulation systems. Among these techniques, Case-Based Reasoning (CBR) allows the implementation of dynamic and adaptive intelligent behaviors, in contrast to other techniques commonly used in the development of computer games, such as Rules Based Reasoning. Although CBR techniques are being exploited in the development of computer games, these techniques are still few exploited in the context of simulation. In this work, a CBR system development framework to be used in the construction of simulators in the Unity graphics engine was plan and implemented. Among other characteristics, this framework allows to build a case base, where each case is formed by typical features of a simulation scenario. The framework also provides different similarity measures to be applied in the new queries. From this memory of problem solving experiences, the framework allows to implement ways to recover similar cases to support the solution of problems. To evaluate the developed framework, an application about the choice of ammunition for an artillery battery was design and implemented. In this application, was constructed a case base with 1000 (one thousand) different cases and, from them, tests involving choice of ammunition were performed, obtaining an accuracy over of 60%.