Utilizing motion capture as an alternative input to promote presence in virtual environments

Abstract

The aim of this project was to explore the feasibility of utilizing motion capture as an alternative control-scheme for games, with particular focus on leveraging it to promote presence in virtual reality environments. This was accomplished by utilizing a variety of classification techniques to classify the data provided by motion capture sensors, using those classifications as replacements for traditional handheld controllers. Multiple means of motion capture and classification were examined, with final implementation settling on the use of Xbox 360 Kinect sensors for capture and a combination of neural networks and dynamic thresholding algorithms for classification. None of the combinations of capture and classification techniques met the minimum framerate of 90 FPS required to be considered an outright success, although the high levels of gesture classification accuracy provided by many of these combinations proved the system to be feasible given the possibility of future work