Pattern Classification of the Grasped Object by the Artificial Hand

This paper describes an a r t i f i c i a l t a c t i l e sense with multi-elements that not only takes the pressure i n f o r mation without crushing an object in the hand, but also discr iminates one pat tern from another on the surfaces being in touch therewi th . The eigenvalue of pat tern of stress d i s t r i b u t i o n on the surface of the receptor elements is derived from the lielmholtz equation. The element spacing of receptor is designed by using the re l a t i ve error of the eigenvalue of the sampled pat tern as the design spec i f icat i o n . A simple experiment c lass i f y ing cy l inders , t r igona l prism and square p i l l a r was performed by the a r t i f i c i a l hand wi th mult i -e lements. As a resu l t , a high percentage of correct answer is obtained,