Intelligent virtual environment using Artificial Neural Networks

Abstract

This paper describes an Intelligent Virtual Environment (IVE) which incorporates Artificial Neural Networks (ANN) in the perception and reasoning of a character in this virtual environment, in order to react intelligently to some given warning signs. First, we explore different types of ANN simulated in MATLAB to understand their operation in order to choose the one that fits to our virtual environment. The environment was created with the UDK game engine and it consists of a character that moves across a working environment to identify warning signals. Later we implemented the Multi-Layer Perceptron (MLP) ANN in this environment. MLP was selected according to data obtained in several tests. This implementation was done by integrating the ANN in the state machines in the source code of the game engine to perform several operations within a controlled work environment.