Haptic perception of shape and hollowness of deformable objects using the Anthrobot-III robot hand

This article presents a methodology for the haptic perception of contour sh apes of almost planar objects grasped by a five-fingered robot hand as well as the detection of any object cavity. The originality of our approach res ides in (1) finding the reaction force patterns at the fingertips of a five -fingered robot hand that grasps different deformable objects (forward prob lem) and (2) using these contact force patterns to find the shapes of grasp ed objects (inverse problem) and (3) to determine material defects such as holes in an object with identified shape. Contact force patterns are genera ted in the forward problem by the finite element method (FEM) and the shape identification in the inverse problem is realized by a supervised neural n etwork architecture using the backpropagation algorithm. Following shape id entification, detection of holes is performed by clustering actual and prot otypical contact force patterns using the self-organizing feature maps of n eural gas networks as an unsupervised hole-screening method. (C) 1999 John Wiley & Sons, Inc.