The geometric model for perceived roughness applies to virtual textures

Abstract

A magnetic levitation haptic device (MLHD) is used to simulate a dithered textured surface made up of conical elements. A constraint surface algorithm allows the cone size, shape, mean inter-element spacing and probe size, to be varied in realtime. The high motion bandwidth, fine spatial resolution and high stiffness of the MLHD produce virtual textures with roughness perception characteristics comparable to geometrically similar real textures. Human subjects use virtual spherical probes of four different sizes to explore textures over a range of element spacings. Roughness magnitude estimates show that roughness initially increases as spacing increases. Maximum roughness is perceived at a spacing governed by the probe-texture geometry. Roughness then decreases as spacing continues to increase. There is an approximately quadratic relationship between texture and spacing. The shape of the magnitude estimation curve, and the texture spacing at which maximum roughness for virtual dithered textures is felt, are similar to those found in real textures having the same geometry. A static geometric model approximately predicts these maxima with consistent underestimation. By incorporating probe velocity into the geometric model, this underestimation can be explained and substantially reduced. Based on these results it is concluded that a haptic device with sufficiently high resolution and bandwidth can be used to accurately generate virtual textures which are perceptually similar in roughness to real textures.