2.2 Agent Architecture
A FSM or GOAP system is just a decision making module, it determines what actions should be carried out and when. However this decision making module is only one part of the agent that resides within the virtual game world and it needs to be able to interact with some, if not all, of the other components of an agent. One of the major tasks in creating any AI system is developing a robust agent architecture to plug the AI system into. The architectures explored were again heavily influenced by Orkin’s work with F.E.A.R. as it is the only real-time planner known to have been developed for games that has any documentation about its architecture publicly available at present.
The GOAP system agent architecture considerations are discussed at length in another of Orkin’s articles (Orkin, 2005). The F.E.A.R. agent architecture employs a modified version of the cognitive architecture as described by the Synthetic Characters Group or C4 from MIT (Burke et al., 2001). The C4 architecture consists of six primary components, the sensory system, a working memory, the action selection system, an internal blackboard, a navigation system and motor system and is represented in figure 2 below.