The use of a new type of neural network (NN) for controlling the trajectory of a robot is discussed. A control system is described which comprises an NN-based controller and a fixed-gain feedback controller. The NN-based controller employs a modified recurrent NN, the weights of which are obtained by. training another NN to identify online the inverse dynamics of the robot. The work has confirmed the superiority of the proposed NN-based control system in rejecting large disturbances.