This paper presents a general recurrent neural network (RNN) model for online control of timevarying robot manipulators. The robot manipulators with the different setting parameters are cooperatively work on an unknown curve tracing. Each joint of the manipulator is respectively provided a learning method to optimize trajectory by training RNN model. In this paper, the proposed RNN model shorten the period of learning and improve the cooperative accuracy than the existing neural networks for solving the problems such as cutting or welding special type of wares. More complicated constructive is to fit for the online cooperation. Simulation results show the effectiveness of this approach, and that the proposed RNN model can successfully learning the inverse dynamics of robot manipulators, perform accurate tracking for a general trajectory.