基于遗传算法求解游戏关卡问题

Using Genetic Algorithm to solve the issue of game points

目前,基于角色扮演的游戏越来越多,为增加趣味性,游戏中存在着很多关卡。为了提高关卡的难度,很多关卡都和迷宫结合在一起,玩家不但要闯迷宫,还要打败BOSS。如何设定一个机器人来实现自动过关,这是一个很具挑战的事。本文基于遗传算法建立一个自动过关模型,该模型对机器人行动方向进行编码,使用多参数评估函数等来实现机器人自动过关,并和情境法进行了对比。实验表明该模型能够有效的解决游戏过关卡问题。

At present, the Role Playing Game is becoming more and more popular. In order to increase the interest of the game, there are quite a few game points in the game. For the purpose of increasing the difficulty of game points, many game points are combined with the maze, the player must not only break through the maze, but also need to defeat the BOSS. How to set a robot to achieve the automatic clearance is a very challenging thing. In this paper, we establish an automatic model based on genetic algorithms. The model encodes the direction of robot actions, the use of multi-parameter assessment of function, such as to achieve the robot automatic clearance. Then we compare the result with situational method. The model experiments show that genetic algorithm is an effective solution to the problem of the game checkpoints.