One Approach to Determine Goals in RTS Games Using Maximization of Resource Production with Local Search and Scheduling

Resources production in RTS games is an area that presents many challenges such as constraint satisfaction, action scheduling and resource management. Despite these challenges, we must consider which resources should be used in a production, since the resources produced characterize a goal that is achieved within the game. There are few studies that consider the choice of the most appropriate resources to compose a goal. Therefore, we developed an new approach that can determine goals within RTS games based on the maximization of resource production. The goals are determined through a quality criterion and the necessary actions for its execution are also determined. The approach uses the Simulated Annealing (SA) algorithm along with techniques that enable its execution within the real time environment. In this paper, emphasis will be given in the description of these techniques. The experiments show that the approach achieves good results operating in real time within the game, as well as competing with experienced human players and other A.I players both in relation to resource production.