Simulating Human Strategic Vision in Real-Time Strategy Games with Holonic Superposition Intelligent Multi-Agent Systems

Article type: Research Article

Authors: Gouws, Gerard^{; *} | Ehlers, Elizabeth Marie

Affiliations: Academy of Computer Science and Engineering, University of Johannesburg, Johannesburg, South Africa

Correspondence: [*] Corresponding author. Email: [email protected].

Abstract: Simulating human-like long-term (strategic) vision in modern real-time strategy (RTS) games remains a challenge. Repetitive artificial behaviour commonly leads to predictable and exploitable predicaments when facing human players. This paper introduces a Real-time Autonomous Superposition Strategy Arena – Ripsaw. Ripsaw is a platform facilitating the participation of autonomous players in an RTS game. Ripsaw attempts to simulate human-like strategic intelligence in the realm of RTS games by incorporating the concept of holonic superposition intelligence. The Holonic Superposition Collaborative Multi-Agent Systems Architecture (Splinter) introduces the concept of holonic superposition intelligence by unifying the Beliefs-Desires-Intentions model, holonic agents and the linear quantum superposition principle. Ripsaw extends Splinter and introduces a framework to simulating human-like strategic vision in RTS games. This paper provides an experiment demonstrating the results of Ripsaw where artificial competitors with differing human-like strategic visions compete. The Ripsaw framework demonstrates the promising approach of incorporating holonic superposition intelligence to simulate human-like strategic vision in RTS games.

Keywords: Beliefs-desires-intentions, linear quantum superposition principle, multi-agent architectures, holonic multi-agent architectures, real-time-strategy games

DOI: 10.3233/JID190005

Journal: Journal of Integrated Design and Process Science, vol. 23, no. 1, pp. 125-140, 2019