|[hawes04phd] Nick Hawes. Anytime Deliberation for Computer Game Agents. Ph.D. thesis, School of Computer Science, University of Birmingham. 2004. [pdf] [bib]|
This thesis presents an approach to generating intelligent behaviour for agents in computer game-like worlds. Designing and implementing such agents is a difficult task because agents in computer games are required to act in real-time and respond immediately to unpredictable changes in their environment. Such requirements have traditionally caused problems for AI techniques. The problem of agent construction is exacerbated by game worlds also being complex environments, and therefore requiring the agents within them to employ powerful (and hence slow) deliberative methods to generate suitable behaviour.
To enable agents to successfully generate intelligent behaviour in such real-time, complex worlds, research has been carried out into two important areas of agent construction. The first of these areas is the method used by the agent to determine future actions in hypothetical situations. To enable an agent to make the most efficient use of the processing time available to it, an action planner has been developed that behaves as an anytime algorithm. The anytime planner is based on a hierarchical task network planner, and provides functionality for the planning agent to interrupt the planning process at any time.
The second area of agent construction that has been researched is the design of agent architectures. This has involved designing and implementing an agent architecture with the necessary functionality to support a planner based on an anytime algorithm. The result of this research is an agent that plays Unreal Tournament and displays behaviour that varies intelligently as it is placed under pressure.
|Download: pdf (968 KB) bib|
|Links: [Google Scholar] [CiteSeer]|