Enhancing handling precision of virtual objects in VR experiences

Abstract

Interaction with virtual objects in a virtual reality environment is still not as realistic as one would expect. When picking objects, users have to make use of external hardware and pre-programmed interaction behaviours. This project investigates how sensors, such as those found in Leap Motion, can be adapted to afford more natural interaction with objects in a virtual space using one’s own hands. A novel interaction technique - state-driven prediction algorithm - has been developed for this research project, building upon the state of the art in terms of hand-detection accuracy. This technique is in turn compared with two other interaction techniques through an empirical study with 27 participants. Results arising from this study show that the novel interaction technique provides a high level of user confidence and a low level of frustration when interacting with the virtual environment. These results are compared with other methods including the Leap Motion’s Interaction Engine.