Computer Vision Laborarory

In this paper, we examine the application of a psychophysical eval-uation technique to quantify the fidelity of haptic rendering meth-ods. The technique is based on multidimensional scaling (MDS) analysis of similarity ratings provided by users comparing pairs of haptically-presented objects. Unbeknownst to the participants, both real and virtual deformable objects were presented to them. In addi-tion, virtual objects were either presented under higher-fidelity ren-dering condition or under lower-fidelity condition in which force filtering and proxy-point filtering were removed. We hypothesized that reducing fidelity of virtual rendering would exaggerate the dif-ference between real and virtual objects. MDS analysis of pair-wise similarity data provided quantitative confirmation that users perceived a clear difference between real and virtual objects in the lower-fidelity, but not in the higher-fidelity condition. In the latter, a single perceptual dimension, corresponding to stiffness, sufficed to explain similarity data, while two perceptual dimensions were needed in the former condition. This study demonstrates how MDS analysis provides an opportunity to visualize and quantify the per-ceptual effects of changes in rendering parameters and how it can be used in the evaluation of haptic rendering scenarios