A Study of Deadzone Structure for Multi-Dimensional Haptic Force Perception

A perceptual deadzone of a reference force, quantified in terms of the just noticeable difference (JND), is a region where any change in the force with respect to the reference force is not perceived. This perceptual deadzone is employed for efficient encoding of the haptic data through an adaptive sampling mechanism. However, for multi-dimensional force stimuli, the effect of the magnitude and direction of the stimuli on the JND has not been investigated yet, and thus, the structure of the deadzone is yet to be determined. This article aims to investigate the perceptual deadzone for 2D haptic force. To achieve this, we designed a psychophysical experiment in which a subject experiences a 2D kinesthetic force stimulus via a force-feedback device and is asked to respond to the change. We employ a machine learning classification-based approach to analyze the \(\textrm{2D}\) perceptual deadzone on collected haptic responses (perceived and non-perceived) of 12 subjects. Our hypothesis is that a classifier optimized on the collected responses, which yields the least prediction error, will define the structure of the deadzone. The characteristics of the optimal deadzone structure have been analyzed in terms of shape, effect of force direction, and symmetry. The results demonstrate that the deadzone is direction variant and radially asymmetric. Thus, both the force magnitude and direction affect force perception. These findings are significant for the fundamental understanding of multi-dimensional force perception by humans, which can be crucial for enhancing haptic force-feedback control systems in virtual reality or robotic controls.