Robotic manipulator trajectory control using neural networks

There are many problems in robotic manipulator control systems due to the increased complexity of the dynamics of robotic manipulators. To overcome these problems, a neural network-based control scheme is described in this paper for robotic manipulators. At present, one of the key problems in designing such a control scheme is how to obtain an efficient training algorithm for learning unknown dynamics. In general cases, however, the approximate models of the controlled robotic manipulators are available. It is believed that the much better learning efficiency of neural networks will be achieved if the prior information can be directly incorporated into the control design. In this paper, the modified conventional backpropagation algorithm and the computed torque method are used for the proposed control scheme.