Intelligent self-learning characters for computer games

In this paper, a novel AI-based animation approach is presented to simulate intelligent self-learning characters for computer games or other interactive virtual reality applications. The complex learning behaviours of the virtual characters are modelled as an evolutionary process so that adaptive AI algorithms such as genetic algorithms have been used to simulate the learning process. The simulation method enables the characters in a computer game environment to have abilities to learn for specific assigned tasks. Its skill for completing the tasks can be developed and evolved through its experiences of performing the tasks. The paper also describes techniques for performance evaluation and optimisation for virtual characters to perform jumping tasks.