Towards an Automatic Determination of Grasping Points through a Machine Vision Approach
Pedro J. Sanz, Jose Manuel Iñesta, A.P. del Pobil · Industrial and Engineering Applications of Artificial Intelligence and Expert Systems · 2022
A system is presented which coordinates a parallel-jaw gripper and a vision system with the aim of choosing grasping points taking into account stability conditions. The choice of grasping points is done by a new heuristic approach that tries to manage all kind of planar parts with or without internal centre of gravity, possibly containing holes. It needs only two moment-based features, centroid and the direction of the main axis, easily obtained from a Freeman chain-code representation of the boundary. The results are exclusively visual parameters, expressed in visual coordinates (pixels). These algorithms have been implemented in a 4D0F robot arm with one camera fixed over the hand. Visual feedback was used in the control system to improve precision up to the limits of the available visual system.