Determination of 3D grasping points using stereo matching and neural network

This paper deals with the synthesis of the determination of 3-dimensional grasping points for unknown object. This is achieved by two steps. First step is to find the whole 3-dimensional geometrical information for unknown object by using a stereo matching. In this step SMW method is employed to find the precision 3-dimensional geometrical information. Second step is to find the 3-dimensional grasping points by using neural network which is trained by grasping data sets resulted from optimization of which objective function is to minimize the force of finger tip and subjected to coulomb friction model with known object. The algorithm is verified by computer simulation.