3D optimal determination of grasping points with whole geometrical modeling for unknown objects

This paper deals with the synthesis of determination of the optimal and automatic three-dimensional grasping points for unknown objects. This is archived by two steps. First step is to find the whole three-dimensional geometrical modeling using stereo matching. In this step, initial process which is partly imported from uncalibrated stereo matching is added to improve the estimation time and to reduce the error rate. Second step is to determine the optimal grasping points using genetic algorithm. In this process, the objective function of optimization minimize the sum of finger's tip force and subjective functions are both static functions and Coulomb friction model. The proposed algorithm is verified by experiment in three-dimensional modeling process and computer simulation in optimization using the known object with different angle. (C) 2003 Elsevier B.V. All rights reserved.