基于决策模型的AI引擎研究与实现
发布时间：2018-08-06 07:53
【摘要】： 随着网络游戏逐渐融入人们的娱乐生活,人们对在网络游戏中的虚拟体验有了更高的要求。以往流行的单纯打怪,过关,升级的模式已不能吸引更多的玩家。网络游戏行业迫切需要更真实,更具有挑战性的对手。我认为,在未来的几年里,AI技术将飞速发展,像Black White和Hello这样的游戏已经让我们为其AI技术而惊叹。游戏玩家正期待出现更多这样的游戏。对于游戏Al引擎也曾有人提出利用遗传算法和神经网络来实现,但是这两种人工智能技术极其复杂,需要占用大量的CPU时间。因此在快速动作的游戏世界,这两种技术并不适用。现在比较流行的实现游戏AI的技术是有限状态机(FSM)。有限状态机是一种基于规则—行动的处理模式,因为它简单、直观、容易修改,所以得到了广泛应用。但是基于有限状态机的AI引擎通常造成游戏角色的非智能行为,比如说：多个非玩家角色(NPC)占用同一路线,造成路径堵塞；怪兽不懂得协同作战等。本文就是在此现状下提出来了一种新的人工智能引擎实现方法—基于决策模型的人工智能引擎。它比传统的基于有限状态机的人工智能引擎更加先进。它使游戏里面的怪兽(MC)和非玩家角色(NPC)能够在不同的情况下做出最有效最理智的决策,使游戏玩家沉浸在一个比较真实的游戏世界里面。 本文结构如下：第一章介绍了游戏引擎基本组成及发展情况,AI引擎在游戏中所处的地位及作用等。第二章讲述传统基于有限状态机的AI引擎设计模式,建立了一个有限状态机的类。第三章讲述本文提出的基于决策模型的AI引擎设计模式及游戏实例。第四章比较两种人工智能游戏引擎的利与弊,着重阐述了基于决策模型的AI引擎的优势所在。第五章通过将把两种AI引擎分别应用到游戏实例中,来比较它们之间的差别。第六章总结全文并对人工智能引擎的发展方向做出了一个预测。
[Abstract]:With the gradual integration of online games into people's entertainment life, people have higher requirements for virtual experience in online games. In the past popular pure fight strange, pass, upgrade the mode can not attract more players. The online game industry urgently needs more real, more challenging opponents. I think that in the next few years, the AI technology will grow rapidly, and games like Black White and Hello have made us marvel at its AI technology. Gamers are expecting more of these games. For the game's Al engine, it has been proposed to use genetic algorithm and neural network, but these two artificial intelligence technologies are extremely complex, which require a lot of CPU time. So in the world of fast action games, these two technologies don't work. The most popular technology to implement game AI is the finite state machine (FSM). Finite state machine (FSM) is a rule-action-based processing model, which is widely used because it is simple, intuitive and easy to modify. But the AI engine based on the finite state machine usually causes the non-intelligent behavior of the game characters, such as: many non-player characters (NPC) occupy the same route, resulting in path blockage; the monster does not know how to cooperate in combat, and so on. In this paper, a new implementation method of artificial intelligence engine based on decision model is proposed. It is more advanced than the traditional artificial intelligence engine based on finite state machine. It enables the game's monster (MC) and the non-player character (NPC) to make the most effective and rational decisions in different situations and immerse the gamer in a more real game world. The structure of this paper is as follows: the first chapter introduces the basic composition and development of game engine and the position and function of AI engine in the game. The second chapter describes the traditional AI engine design pattern based on finite state machine, and establishes a class of finite state machine. The third chapter describes the AI engine design pattern and game examples based on decision model. Chapter 4 compares the advantages and disadvantages of the two artificial intelligence game engines, focusing on the advantages of AI engine based on decision model. Chapter 5 compares the two AI engines by applying them to game instances. The sixth chapter summarizes the full text and makes a prediction to the development direction of artificial intelligence engine.
【学位授予单位】：暨南大学
【学位级别】：硕士
【学位授予年份】：2010
【分类号】：TP391.3
本文编号：2167022
[Abstract]:With the gradual integration of online games into people's entertainment life, people have higher requirements for virtual experience in online games. In the past popular pure fight strange, pass, upgrade the mode can not attract more players. The online game industry urgently needs more real, more challenging opponents. I think that in the next few years, the AI technology will grow rapidly, and games like Black White and Hello have made us marvel at its AI technology. Gamers are expecting more of these games. For the game's Al engine, it has been proposed to use genetic algorithm and neural network, but these two artificial intelligence technologies are extremely complex, which require a lot of CPU time. So in the world of fast action games, these two technologies don't work. The most popular technology to implement game AI is the finite state machine (FSM). Finite state machine (FSM) is a rule-action-based processing model, which is widely used because it is simple, intuitive and easy to modify. But the AI engine based on the finite state machine usually causes the non-intelligent behavior of the game characters, such as: many non-player characters (NPC) occupy the same route, resulting in path blockage; the monster does not know how to cooperate in combat, and so on. In this paper, a new implementation method of artificial intelligence engine based on decision model is proposed. It is more advanced than the traditional artificial intelligence engine based on finite state machine. It enables the game's monster (MC) and the non-player character (NPC) to make the most effective and rational decisions in different situations and immerse the gamer in a more real game world. The structure of this paper is as follows: the first chapter introduces the basic composition and development of game engine and the position and function of AI engine in the game. The second chapter describes the traditional AI engine design pattern based on finite state machine, and establishes a class of finite state machine. The third chapter describes the AI engine design pattern and game examples based on decision model. Chapter 4 compares the advantages and disadvantages of the two artificial intelligence game engines, focusing on the advantages of AI engine based on decision model. Chapter 5 compares the two AI engines by applying them to game instances. The sixth chapter summarizes the full text and makes a prediction to the development direction of artificial intelligence engine.
【学位授予单位】：暨南大学
【学位级别】：硕士
【学位授予年份】：2010
【分类号】：TP391.3
【引证文献】
相关硕士学位论文 前1条
1 张时铭;短道速滑仿真系统中智能体决策过程的研究与实现[D];哈尔滨工业大学;2011年
，本文编号：2167022
本文链接：https://www.wllwen.com/kejilunwen/rengongzhinen/2167022.html
