Data-based models for deformable objects: Sensing, acquisition, and interactive playback

Many haptics-enabled virtual environments require the inclusion of deformable solids for increased realism. For example, some medical procedure simulations require deformable organs. The feel of an organ can indicate the location of critical underlying structures or the presence of disease. The models currently used in these simulations tend to be created by hand in a time consuming manner and/or are based on expensive computational methods (e.g. finite elements).; This thesis describes the design and implementation of an automated system to build data-based models of elastic deformable objects and to render these models in an interactive virtual environment. By automating model creation, more realistic models in greater numbers can be included in these interactive simulations.; The system consists of two components: a data acquisition system and a model creation and rendering system. In the data acquisition system, the geometry of an object is sensed by a novel structured-light sensor. The sensor is designed to recover depth from single images which allows it to work on fast moving and deforming objects. The behavior of the object and haptic information is captured through active probing using a haptic device as a robot. Techniques for contact detection, slip detection on deformable surfaces, and enhanced open-loop force control enable the robot to explore the force field at various contact locations on the object's surface.; In the model creation and rendering system, measurements are transformed into a data-based model and efficiently rendered in an interactive haptic environment. Discrete samples are interpolated into a continuous force field, and friction properties are extracted. The running time of a rendering loop iteration scales logarithmically to large data sets and is empirically shown to be less than 50 mus for test data sets with measurements from about one hundred contact locations.; The system has been implemented on hardware readily available to haptics researchers and demonstrated with nonlinear inhomogeneous test objects.