A Neural Network Approach for optimal grasp planning

Abstract

In this paper, the Neural Network (NN) Approach is used to find the best point on the object, for executing object manipulation task by a manipulator. The MAG performance index is calculated for some sample points of objects heuristically by MSC.ADAMS and MATLAB co-simulation for the 6DOF Stäubli <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">©</sup> TX40 arm. These samples then would be used to train a feed-forward back propagation neural networks. The result is the dynamics model of the robot and the grasped object in which the MAG performance index value is the input and the position of the best grasping point of the objects which maximizes the MAG index is the output.