MODELING THE BEHAVIOR OF NON-PLAY CHARACTERS BASED ON ARTIFICIAL INTELLIGENCE

Abstract

Computer and video games are one of the largest segments of the entertainment industry, which has been developing rapidly in recent years. Even the pandemic benefited this industry. The variety of games is impressive, users want to get high-quality diverse content, an important part of which are non-game characters. A Non-Player Character is a character controlled by a program or wizard. Non-Player Characters (friendly or hostile to the player) serve as an important means of creating a game atmosphere, they motivate players to take certain actions, and are the main source of information about the game world and the plot of the game. Modeling the actions of Non-Player Characters is an important task in game development to improve game quality and enhance player satisfaction. In modern game development, artificial intelligence algorithms are used to solve this problem, including machine learning, for example, to create «smart» bots that will fight with the player. Game AI is not capable of thinking or creativity, its actions are determined by developers, it adapts to the situation and changes behavior depending on the context. The main approaches in the development of the behavior of non-game characters are the creation of own solutions for modeling the behavior of characters based on the basic concepts of AI. This work is devoted to modeling the behavior of Non-Player Character using artificial intelligence algorithms, in particular the Behavior Tree and the A* search algorithm. A Behavior Tree allows a Non-Player Character to make decisions depending on the state of the environment and react to its changes. Movement is an important part of the behavior of Non-Player Characters, which are often not controlled by the player, but often game logic requires movement by the shortest route for the sake of game dynamics. The integration of the simulated behavior into the already existing free models of the Unity game engine was carried out, the correct work of finding the shortest path according to the A* algorithm, taking into account various cases of the environment and the character itself, as well as the functionality of the navigation system between scenes was proven. The developed behavior models can be further integrated into existing models or used in the creation of new games and characters.