Modeling Three-Dimensional Interaction Tasks for Desktop Virtual Reality

A virtual environment is an interactive, head-referenced computer display that gives a user the illusion of presence in real or imaginary worlds. Two most significant differences between a virtual environment and a more traditional interactive 3D computer graphics system are the extent of the user's sense of presence and the level of user participation that can be obtained in the virtual environment. Over the years, advances in computer display hardware and software have substantially progressed the realism of computer-generated images, which dramatically enhanced user’s sense of presence in virtual environments. Unfortunately, such progress of user’s interaction with a virtual environment has not been observed. The scope of the thesis lies in the study of human-computer interaction that occurs in a desktop virtual environment. The objective is to develop/verify 3D interaction models that can be used to quantitatively describe users’ performance for 3D pointing, steering and object pursuit tasks and through the analysis of the interaction models and experimental results to gain a better understanding of users’ movements in the virtual environment. The approach applied throughout the thesis is a modeling methodology that is composed of three procedures, including identifying the variables involved for modeling a 3D interaction task, formulating and verifying the interaction model through user studies and statistical analysis, and applying the model to the evaluation of interaction techniques and input devices and gaining an insight into users’ movements in the virtual environment. In the study of 3D pointing tasks, a two-component model is used to break the tasks into a ballistic phase and a correction phase, and comparison is made between the real-world and virtual-world tasks in each phase. The results indicate that temporal differences arise in both phases, but the difference is significantly greater in the correction phase. This finding inspires us to design a methodology with two-component model and Fitts’ law, which decomposes a pointing task into the ballistic and correction phase and decreases the index of the difficulty of the task during the correction phase. The methodology allows for the development and evaluation of interaction techniques for 3D pointing tasks. For 3D steering tasks, the steering law, which was proposed to model 2D steering tasks, is adapted to 3D tasks by introducing three additional variables, i.e., path curvature, orientation and haptic feedback. The new model suggests that a 3D ball-and-tunnel steering movement consists of a series of small and jerky sub-movements that are similar to the ballistic/correction movements observed in the pointing movements. An interaction model is originally proposed and empirically verified for 3D object pursuit tasks, making use of Stevens’ power law. The results indicate that the power law can be used to model all three common interaction tasks, which may serve as a general law for modeling interaction tasks, and also provides a way to quantitatively compare the tasks.

• View at Publisher
• View at Worldcat

Free Full Text ( Author Manuscript , 16mb )