Information Technology and Security, Vol. 9, Iss. 2 (17)

Постійне посилання зібрання

https://ela.kpi.ua/handle/123456789/54423

Переглянути

Перегляд Information Technology and Security, Vol. 9, Iss. 2 (17) за Ключові слова "artificial intelligence"

Зараз показуємо 1 - 1 з 1
  • Ескіз
    ДокументВідкритий доступ
    Methods and approaches analysis of artificial intelligence designing for real time strategy game
    (Institute of Special Communication and Information Protection of National Technical University of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute”, 2021) Dorohyi, Yaroslav; Svyrydenko, Oleksandr
    The research provides a detailed analysis of approaches to creating AI in video games. The main area of research is AI for real-time strategies, as this genre is characterized by the complexity of the game environment and the practice of creating a comprehensive AI, consisting of several agents responsible for a particular aspect of the game. The analysis shows that the main areas of use of AI methods in strategies are strategic and tactical decisions, as well as analysis of the current situation and forecasting the enemy and his chosen strategy. Among the analyzed approaches to tactical AI, reinforcement, game tree search, Bayesian model, precedent-based solutions and neural networks are most often used. Popular approaches to building strategic AI are precedent-based decision-making, hierarchical planning, and autonomousachievement of goals. When creating a module for research and determination of plans, the most popular methods are deductive, abduction, probabilistic and precedent. In addition to the considered methods, others are used in the development, but they are not as popular as above, due to problems with speed or specific implementation, which does not allow to adapt them to the standard rules of genre games. Comparison of algorithms and implementations of AI in the framework of commercial and scientific developments. Among the main differences are the high cost of commercial development of complex agents, as well as the specifics of the scientific approach, which aims to create the most effective agent in terms of game quality, rather than maximizing positive impressions of players, which is the basis of commercial development. The reasons for insufficiently active development of scientific research in the field of AI for games in general and the genre of real-time strategies in particular are described.