A Neural Gripper for Arbitrary Object Grasping

Abstract

This paper presents a two-stage neural system to determine the contact points between a three-® ngered gripper and an object of arbitrary shape. In the ® rst stage, a CCD camera captures the image of the object and such an image is transformed into a two-dimensional outline through a nearest neighbour algorithm. In the second phase, two neural networks, functioning in cascade, select three contact points in the outline. A competitive Hop® eld neural network de® nes an approximate polygon considering a reduced number of boundary points of the original outline. Then, a supervised neural network, either a multi-layer perceptron or a radial basis function (RB F) network, ® nd the contact points. The experiments suggest that the RB F network trained by the global ridge regression method is suitable for on-line applications and presents the best overall performance in terms of accuracy and robustness to noise. Moreover, this method is able to ® nd correctly the contact points for objects of arbitrar y shapes.

A self-organizing state trajectory planner applied to an anthropomorphic robot hand

“…This material is lighter than the aluminum versions used in the first [16] and second [17] hand/gripper generation.…”

“…In this work a new concept of an anthropomorphic robot hand is presented. The prototype, called Kanguera (ancient indigenous word for bones outside the body) Fig.1(b), is the fourth generation of hands/grippers developed at the EESC -USP Mechatronics Lab [16][17] [18] and it presents significant performance improvements when compared to the previous systems.…”