Realistic Agent Movement in Dynamic Game Environments

One of the greatest challenges in the design of realistic Artificial Intelligence (AI) in computer games is agent movement. Pathfinding strategies are usually employed as the core of any AI movement system. This paper examines pathfinding algorithms used presently in games and details their shortcomings. These shortcomings are particularly apparent when pathfinding must be carried out in real-time in dynamic environments. This paper proposes a strategy by which machine learning techniques such as Artificial Neural Networks and Genetic Algorithms can be used to enhance traditional pathfinding algorithms to solve the real-time aspect of this problem. We describe a test bed system, currently in development, that incorporates these machine learning techniques into a 3D game engine.