Haptic deformation using graphics hardware and kd-trees

In this paper, we describe a haptic deformation algorithm using graphics hardware and spatial partitioning data structure. Since object-based collision detection methods are generally dependent on the geometrical structure of objects, these are not fast enough for deformable objects. Thus Z-buffer depth comparison using graphics hardware is used for detecting collision between a haptic probe and an object. In order to determine what region to deform in response to a user’s actions, we adopted a nearest neighbor search algorithm based on a kd-tree structure [Moore. 1991]. At the same time, the reaction force to be presented to the user is calculated in the haptic rendering loop. To compute response force for object’s deformation, mass-spring methods are utilized. Where masses are assigned to vertices and a set of spring are allocated to connect vertices.