Neural Network-based H Control for Fully Actuated and Underactuated Cooperative Manipulators

This paper develops an H_infin control based on neural networks for fully actuated and underactuated cooperative manipulators. The neural networks proposed in this paper adapt only the uncertain dynamics of the robot manipulators, actuating as a complement of the nominal model. The H_infin performance index includes the position errors as well the squeeze force errors between the manipulators end-effectors and the object, which represents a complete disturbance rejection scenario. For the underactuated case, the squeeze force control problem is more difficult to solve due to the lost of some degrees of actuation of the manipulators. This problem is addressed and a practical solution is found. Results obtained from an actual cooperative manipulator, which is able to work as a fully actuated and an underactuated manipulator, are presented.