Creating Human-like Behavior in Game Agents

Assigned Computer Science Games Engineering Bachelor Master Project Thesis Games Research Lab

Background

Many games require interaction between the player and other, non-player characters (NPCs), which are controlled by an artificial intelligence (AI). NPCs sometimes also occur as substitutes for human player in multiplayer games. In both situations, an adequate level of AI is needed to achieve a satisfactory gaming experience. Thus, in many cases the AI must not have super-human playing skills but be on par with the players to provide a realistic experience. Machine learning methods like reinforcement learning and imitation learning make it possible to teach AI this desirable behavior.

Tasks

• Research commonly used methods in reinforcement learning and imitation learning
• Find or create a suitable game environment to train AI agents
• Use this game environment to create NPCs with human-like behavior by means of machine learning
• Evaluate the resulting behavior by means of a user study

Contact Persons at the University Würzburg