Data-Driven Modeling of Anisotropic Haptic Textures: Data Segmentation and Interpolation

This paper presents a new data-driven approach for modeling haptic responses of textured surfaces with homogeneous anisotropic grain. The approach assumes unconstrained tool-surface interaction with a rigid tool for collecting data during modeling. The directionality of the texture is incorporated in modeling by including 2 dimensional velocity vector of user's movement as an input for the data interpolation model. In order to handle increased dimensionality of the input, improved input-data-space-based segmentation algorithm is introduced, which ensures evenly distributed and correctly segmented samples for interpolation model building. In addition, new Radial Basis Function Network is employed as interpolation model, allowing more general and flexible data-driven modeling framework. The estimation accuracy of the approach is evaluated through cross-validation in spectral domain using 8 real surfaces with anisotropic texture.