Higher-Order Experimental Haptic Force Models of Mechanical Devices Using Moving Ridge Regression

Accurate modeling of physical devices is one of the keys to improving the realism of haptic simulations. This paper presents general, locally linear haptic force models to describe the feel of complex mechanical systems that exhibit nonlinear static and dynamic behavior. The parameters of these models are estimated from experimental data using moving ridge regression. Nonlinear variations of the locally linear model are presented and analyzed, and the goodness-of-fit of these models is compared. Initial results suggest that higher-order terms do little to improve the quality of this class of haptic models, and that reduced-order models should be further investigated.