Experimental validation of a tactile sensor model for a robotic hand

We describe a tactile sensor for a robotic hand, based on the mechanoreceptors in the glabrous skin of the human hand to replicate the sensory function of both slow adapting and fast adapting receptors. Strain gauges are used for the slow adapting receptors, and polyvinylidene fluoride (PVDF) film was used to replicate the function of the fast adapting receptors. One unit sensor consisted of four strain gauges and a single PVDF film, embedded beneath a square protrusion. The protrusion helped localize the applied force onto the region or `receptive field' of the sensing unit. Strain gauges were orientated to enable the unit sensor to identify the tri-axial force components. Multiple linear regression was used to predict the components of force. The regression model with interaction terms gave good prediction with mean percentage errors of less than 15% for each force component.