Grasp planning for a two-fingered parallel-jawed gripper

The main emphasis of this paper is on grasp planning for a two-fingered parallel jawed gripper (typically used in industries) utilizing visual feedback for improved grasping for a planar object. Appropriate placement of fingers on the object plays an important role in grasp stability. This task is relatively easier for a priori-known target. But when the model of the object is not known a priori, it becomes difficult to plan the grasp and the usability of manipulator becomes limited. The main difficulty is in deciding upon the grasp points in a random object at a random location. A general approach has been presented in this paper which is applicable to any planar object. The approach uses the 2D features extracted from the image such as centroid, dimensions and principal axes to compute the grasp points' coordinates and the location of grasp axis for stability. A specialized approach is also presented which deals with objects containing only straight edges. The specialized approach utilizes Hough transform and image features in conjunction to enhance the performance of the algorithm. The proposed method can handle abrupt changes in the contour of an object as the algorithm takes object corners into account. Experiments were conducted in the laboratory to test the effectiveness of the proposed algorithm and validate the results. Copyright © 2006 by ASME.

Duke Scholars

Author Devendra P. Garg Thomas Lord Department of Mechanical Engineering and Materia ...