Home About MOVIII Research Demonstrator projects Publications Events Contacts

Haptic Interaction with Dynamic Data

Karljohan Lundin and Anders Ynnerman

Haptic interaction with volumetric data is an area within which the NVIS group, part of the MOVIII consortium, has significant experience. The group has recently developed the volumetric haptics toolkit, VHTK, built upon the haptics interface (H3D), and which can be obtained from www.volumehaptics.org. The current implementation is designed around interaction with static volumetric data and provides methods for constructing complex haptic modes to perceive such static properties as density, viscosity, friction, flow velocity and vorticity. The haptic interaction has been found to provide a great assistance to the user, both by providing an additional channel to include physical properites information which are hard to perceive when displayed visually and by providing guidance for the user in interacting with and interpreting the visual information.

 Current work with haptics in MOVIII involves the inclusion of interaction with time-varying data to allow the inclusion of movement of the data and changes within the data volume itself while interacting. This approach has initially been targetted at dynamic data such as the results of scientific simulations and time-based medical scans and has been found to be equally as effective with this new form of data. The new approaches we have developed allow, for example, the user to feel the changes in the data as fluid flows change in response to modifications of an aircraft model or to feel the beating of a human heart and how the blood flow through the associated vessels changes as the heart goes through its cycle.  

To give the user full interaction with the volumetric data, however, the haptic simulation must also take account of the users interactions in more direct ways. Rather than simply changing the physical forces being presented to the user the haptic simulation itself muct allow the users activities to change the data itself. This is particularly important where the user is palpating the data to gain insight into how the volumetric data reacts. A very clear example of this kind of operation is in medical diagnosis where the application of pressure to a tissue results in different responses depending on the nature and condition of the tissue. Doctors 'feel' for imflammation, sub-dermal fluid retention, torn and damaged muscles and broken bones and it is the response of the tissue to pressure that provides the doctor with the information they require. To capture this behaviour requires not only that the volumetric representation of the data be of high quality, but that the data itself responds to the interaction.

This approach will also have important applications in many other disciplines since it ties in strongly with ideas in computational steering but the initial focus will be on the medical side, dealing with the dynamical modelling of tissues using methods from engineering (finite element modelling) and the complex parameterization techniques (such as those applied in System Identification) to decude the problem to one which can be solved at the extremely high update rates needed for the haptic interaction scheme.