Study On The Natural Interaction Of Flexible Virtual Hand

Virtual hand interaction technique is widely used in applications like virtual assembly, medical simulation, entertainment and animation etc.,. Therefore, the real-time capability and effectiveness in interaction are primarily demanded for virtual hands, as well as realistic performance. However, due to the flexibility and complexity of hand movement, current technique can not satisfy those requirements in interaction experience, because of severe reality deficiency or effectless interaction.Here, we conduct this study to explore an interaction mode which can achieve both appearance reality and real-time capability during virtual hands operation. In virtual hand interaction, information from hand is transferred into virtual environment. After processing, results are shown to users in graphical display. Fluency, fidelity and effectiveness can be used to judge merits of the interaction process. It makes the task much more complicated to keep balance between visual/motion realism and real-time performance along with limitation from hardware devices.Based on the analysis of interaction requirements and computational efficiency for flexible virtual hands, a natural framework for flexible interaction is established for forward/backward kinematics interaction in this thesis. This framework incorporates dual quaternion into modeling process to drive the geometrical deformation of virtual hand, maintaining the object shape during three-dimensional movement and keeping high computational efficiency. Subdivision srface method is utilized to construct realistic appearance, and also virtual hand movement structure for real-time interaction is simply modeled based on anatomy of real hand. Meanwhile, collision detection technique is introduced for real-time interactive tasks.Forward kinematics driven interactive framework is tested with respect to hand gesture interaction and operative tasks. Experiments demonstrate the effectivieness of using this model for interaction, satisfying design requirements.Taking the price of datagolve and movement constraint caused by wearable hardware into consideration, a much more efficient and cost effective interaction mode is proposed here to complete the virtual hands’operational tasks. Research outcome about human hand grasping from psychology is introduced here to generate the fingertip’s movement trajectory for virtual hands. Actually, it is the automatic grasping algorithm along with judgment criteria, rather than datagloves with finger joint angle, that helps to achieve automatic and stable grasping performance.To evaluate and compare the grasping performance of dateglove based forward movement interaction mode and grasping trajectory based automatic grasping algorithm, user experiment is finally conducted in this dissertation. Taking hard grasping task for example, the performance is then observed quantitatively and qualitatively. Quantitative indicators include operational efficiency, user operation efficiency; while qualitative ones are mainly user’s feelings, which are also quantified by questionnaire. The results show that automatic grasping virtual hands over take dataglove driven ones in computational and user operational efficiency, also better than latter with respect to user experience.