Evolving Coordination for Real-Time Strategy Games

A Systematic Review of Coevolution in Real-Time Strategy Games

“…The corresponding evolutionary algorithm (EA) will evolve a counter strategy to improve the situation of this species in the game. Such behaviour keeps going and pushes both species to improve their own strategies yielding what is called an arms race [1], [26], [31], [34], [75]. The successful CA seeks to preserve such behaviour for the longest time possible.…”

“…It is worth mentioning that two of the games used in the reviewed papers were turn based, rather than actual RTS games. Bellus Bellum Gratia (BBG) was used in [33], [34], and Planet Wars, the Google AI challenge introduced in 2010, was used in [18], [25], [53]. These papers are still considered in this review as the games used here were adjusted and equipped with features that made them closer to RTS games than TBS games, as will be explained later.…”

“…They reported that attack coordination is more effective in producing robust strategies than spread coordination, and also that it is easier to evolve better attack coordination in coevolution using a single environment compared to using multiple ones. Two years later, the work was extended in [34] by creating a coordination profile for each strategy (spread/attack coordination managed separately). These profiles were analysed throughout the coevolution process to monitor changes in the coordination and importance of such changes.…”

“…EAs in CAs: Table 6 shows that researchers prefer to coevolve using GAs rather than GPs, yet GP was used more frequently than the other EAs. In a further investigation of [42], [45], [75], [46], [66], [65], [40], [67], [1] Micro Collect resources [30], [60], [46], [37], [44], [38], [22], [19] Create buildings/units [30], [46], [33], [34] Create build orders [10], [9], [11], [12] these two EAs, we found that GA was used in 100% of the publications that used coevolution to evolve game contents (five publications). Moreover, GA was also used in 10 of the 12 publications that used coevolution to evolve NPCs (83.33%).…”

“…[37], [15], [52], [19], [51], [10], [9], [25], [66], [11], [18], [12], [53], [67] [63], [42], [75], [33], [34] GP [44] Neural Networks vs Genetic Programming vs Modular controller [73] game strategies. We do not have an explanation for this; however, maybe the lack of information on the used memory technique hinders the appearance of using this technique for other coevolution purposes.…”