A NEURAL-NETWORK ASSISTED COMPUTED TORQUE METHOD FOR MANIPULATOR TRACKING CONTROL-PROBLEMS

Based on the structure of the computed torque method, this paper devel ops a neural control method for manipulator trajectory tracking proble ms. Compared with conventional approaches, the proposed method has the following novel features. First, by using a neural network as the pla nt model, the proposed method can handle structural as well as unstruc tural uncertainties. Secondly, with a simple compensation scheme, the proposed approach can effectively avoid tracking performance degradati on caused by unmodelled dynamics and/or neural network learning errors . Computer simulation results are given to demonstrate the feasibility of the proposed control method.