Visual Servoing Control Based on Fuzzy Behavior and Neural Networks

This paper proposed a visual servoing control scheme for positioning a robot manipulator. Similar to the behaviour of human, this scheme is divided into two steps: firstly a fuzzy logic-based visual servoing controller is used to guide the gripper into the neighbourhood of the object, and then one local neural network is applied to position the gripper in a desired pose. The whole control needs no calibration of the robot and the cameras and can utilize humans´ experience and has high precision. Simulation results on a 6 DOF industrial robot show the validity and effectiveness of the proposed control scheme.