Synergistic Visual / Haptic Computer Interfaces 1

While graphical visualization has advanced our ability to understand large multi-dimensional data sets, some types of data are still diicult to convey visually. Combined visual/haptic computer interfaces may allow users to explore multi-dimensional data sets more naturally, where the human haptic sense involving touch, limb position, and muscle tension provides a complementary information channel to convey certain data properties. We discuss the development of haptic rendering modes and combined visual/haptic rendering modes which can convey complex, multi-dimensional data. Tests to better understand human haptic and visual/haptic perception of the new data rendering elements are also outlined and initial results are presented. 1. Introduction Human vision is well-suited for 2-dimensional (2D) pattern understanding, and for identifying solid objects in 3-dimensional (3D) spaces. It is, however, less adept at understanding the following types of data: scalar elds on 3D domains, e.g., temperature in a volume, vector elds, e.g., uid ow velocity throughout a volume, and tensor elds, e.g., stress tensors in elastic solids. Traditional visualization techniques attempt to reduce these problems to the display of 2D patterns through techniques such as slicing, coloring, contouring, iso-surfacing, and streamlining. In cases where these dimension-reduction techniques fail, such as in ten-sor eld rendering where the properties of matrices at each point are of interest, symbolic representations (glyphs) are often used. Figure 1 shows an example of the type of visualization diiculties that are often encountered. Here, cones are plotted to indicate the magnitude and direction of the electromagnetic eld on and around a large gap Lorentz force actuator. Visual clutter severely limits understanding of the variation of the eld in the volume. Haptic perception, in contrast, is rather poor at determining global shape or spatial patterns. However, it has complementary capabilities in discerning local Figure 1: Some visualization problems are diicult for purely visual interfaces. (Data and gure provided by Jay St. Pierre, actuator manufactured by Left Hand Design Corporation.) properties, such as surface curvature, vector eld direction , material viscosity, and stiiness. For example, it would seem that presentation of the eld in Figure 1 as a force vector varying through the volume would be very natural to understand haptically, and this may relieve some of the visual clutter by only visually rendering the eld on the surface of the actuator. Yet haptic perception modalities for computer interface are poorly understood and essentially unused. This is largely because, until recently, no high-delity haptic interfaces …