Joint controller for the object-pose controlling on multifinger grippers

A human hand as a very complex grasping tool can handle objects of different sizes and shapes. Many research works have been done to develop artificial robot hands with similar capabilities to a human hand, such a robot hand is also called a multifinger gripper. Most parts of this research are dedicated to control of multifinger grippers with emphasis on the finger tips or finger joints. By controlling a multifinger gripper, we enable the gripper to handle an object; in another words, controlling a multifinger gripper can be viewed in terms of controlling an object's pose and the forces between the object and its environment. Hence, an object-pose controller with feedback from an object pose sensor suits multifinger gripper control. Also due to the non-linear dynamic system behavior in the joints of most multifinger grippers, an effective, easily-adaptable joint controller is employed. The paper discusses the object pose controller with great detail in a new joint controller. Since the joint controller is based on fuzzy and neural-network algorithms, we do not use an exact analytical model for this case.