Adaptive neurofuzzy control of a robotic gripper with external disturbances

Abstract

Robotic grippers are commonly required to grasp and manipulate loads under a wide range of conditions, without the from the end effector, and avoiding damage to the load. We have previously demonstrated the adaptive neurofuzzy control of a simple gripper with a two-input, one-output action. When the robotic gripper is integrated with a multi-degree of freedom robot, the controller could face the curse of dimensionality. To show that satisfactory control was still feasible, we undertook extensive simulations of this gripper mounted on a multi-degree of freedom robot. We also report enhanced control by using the gripper's acceleration as an additional input.