A Visual Servoing Control for Robot Based on Fuzzy Behavior and Neural Networks

A new visual servoing control schemes is proposed for the robot arm movements, which is similar to the pick behavior of human. Firstly a visual servoing controller based on fuzzy logic is used to move the gripper into the neighborhood of the object. Then one local neural network uses image error to position and orientate the gripper exactly. The whole control, which needs no calibration of the robot and the cameras and can utilize humans´ experience, has fast speed and high precision. Simulation results show the validity and effectiveness of the proposed control scheme.