First person shooter (FPS) games, online card games, and massively multiplayer online playing games (MMORPGs) are popular and extremely diverse online game genres. Our research in this paper is based on one of the most vibrantly used FPS games, the Counter-Strike (CS). Encouraged by game Artificial Intelligence (AI), we have incorporated computer game bots for our simulation of the CS game. Computer game bots are non-player characters (NPCs) that simulate human game players and display human-like behavior while playing against humans. We propose a state-of-the-art method towards game design using a game theory based learning algorithm called Fictitious play. The bots (non-player characters), in this method, assume that the opponents are playing a fixed strategy and hence based on their past experiences they plan their next move. This research explores how a computer game bot adapts to the dynamic human player0s behavior while playing the simulated game of CS, thereby leading to its unpredictable behavior. Here we focus only on an individual agent at a time when it comes to experimentation.
Game theory concepts in AI can have an important role to play in the implementation of numerous computer game strategies that can provide a number of mathematical tools to understand the possible opponent strategies. This not only provides the game developers with a useful tool for implementing various complex strategies for the computer games, but can also contribute towards their marketing value. One such learning algorithm in game theory is called Fictitious play (FP). Fictitious play was introduced as a solution to find Nash equilibrium. An agent using fictitious play considers that opponents are playing a fixed strategy and keeps track of the strategy being played by them. Based on these observations, agents execute their future actions. Fictitious play is a simple and efficient algorithm. Fictitious play being a light weight algorithm can be used in internet based virtual environment for three-dimensional games. This paper explores the use of fictitious play in game AI, specifically for computer game bots.