Development of a behavior pattern for a game bot with artificial intelligence elements based on Markov chains
Abstract
Development of a behavior pattern for a game bot with artificial intelligence elements based on Markov chains
Incoming article date: 11.02.2024The subject of this article is the development of a behavior pattern with AI elements for an opponent bot in the single-player game Steal Tower. The essence of which is to collect resources to build a tower faster than opponents. To create the illusion that the same people are playing against the player, an imitation stochastic model based on the Monte Carlo method for Markov chains has been developed. Based on the results of its tests, balanced system parameters were determined, which are embedded in the behavioral pattern of the bot, which is implemented using the Enum AIStates enumeration consisting of five states: Idle (inactivity), GoTo (movement) and GoToWarehouse (return to the warehouse), Win (victory), Loose (scoring). Each of them has developed functions for the optimal behavior of the bot given in the article. So for the GoTo state, functions have been created that analyze the benefits of different types of behavior: steal or collect, or walk to the warehouse or to the tower.
Keywords: game intelligence, behavioral pattern, live world emulation, bot behavior scenario, state structure, Markov chains, Monte Carlo method, simulation model, Unity environment, C# language