By any measure, the human hands and brain form a remarkable system for perception and manipulation. Of particular interest to us is the ability to perform tactual stereognosis: i.e., to identify objects by touch. In a landmark study, Lederman and Klatzky (1987) showed that stereognosis is subserved by a set of stereotypical “exploratory procedures” (EP) such as enclosure for shape, static contact for temperature, lateral motion for texture, unsupported holding for weight, etc. Moreover, they showed that EPs are accessed hierarchically because they differ in terms of their duration and breadth of sufficiency for identification. Lederman and Klatzky proposed that these findings could form the basis of a computational theory of haptics. Our approach to this project will be deeply integrative. For example, we do not see it as adequate to add an artificial skin to existing hand hardware. Instead, it is necessary to carefully plan and integrate the mechanical behavior of the hand, the mechanical and sensing properties of the skin, and the structure of the software control. In particular, we are guided by four organizing principles: 1) Compliance is essential, 2) Prior information is essential, 3) Hierarchy of Exploratory Procedures, and 4) Tradeoff of Spatial and Temporal Frequencies.
Keywords: Haptic Hand, Tactile Object Recognition, Skin Sensor, Robotic Hand, Stereognosis
