Agent-based simulated cognitive intelligence in a real-time first-person entertainment-based artificial environment

The D'Artagnan Cognitive Architecture (DCA) sets forth to provide an exploration into one possible approach and solution for creating an agent-based Autonomous Intelligent Platform (AIP) for controlling a Computer Generated Force (CGF). The Cognitive-based Agent Management System (CAMS), an attention focus based Multi-Agent System (MAS) architecture that emphasizes communication link and dynamic agent control in reconfigurable topologies, is used to realize the DCA. We present the methodology, design, and basic implementation of CAMS-DCA in the Quake II environment against a collected set of twenty human player trials over a developed reference set of one hundred unique and increasingly difficult levels. Understanding human and agent performance through a set of performance and clustering metrics, we show that a basic implementation of our system was able to complete 29% of the reference level set of which 73% were within human performance levels and 15.4% were within our definition of human-consistency.