Precise and realistic grasping and manipulation in Virtual Reality without force feedback

摘要

This paper introduces a physically-based approach of grasping and manipulation regarding virtual objects that would enable fine and stable grasping without haptic force feedback. The main contribution is to enhance an existing method which couples a virtual kinematic hand with a visual hand tracking system. The mismatches between the tracked and virtual hands often yield unstable grasps, especially for small objects. This is overcome by the implementation of grasping assistance based on virtual springs between the tracked and virtual hands. The assistance is triggered based on an analysis of usual grasping criteria, to determine whether a grasp is feasible or not. The proposed method has been validated in a supervised experiment which showed that our assistance improves speed and accuracy for a "pick and place" task involving an exhaustive object set, sized for precision grasp. Moreover, users’ feedback shows a clear preference for the present approach in terms of naturalness and efficiency.