Moment invariant-based multi-target recognition and grasping for robot manipulator

A moment Invariant-Based multi-target recognition and grasping method aimed to the visual servoing control for robot manipulator is presented. The recognizing for the assigned target objective among the multi-targets is realized by using the image region segment technique, features set composed of 7 moment invariants and Euclidean distance measurement. The grasping recognized target objective is completed with the method of position-based visual servoing control. The experimental study is done on a GB-400 4 DOFs robot manipulator by using the proposed method and its effectiveness is verified.