Robust adaptive neural network tracking control for manipulators with unmodeled dynamics

Unmodeled dynamics are the unavoidable nonlinear effect that can limit control performance in robotic systems. The unmodeled dynamics of the system include uncertainty or unknown and unmeasured states. Meanwhile, it is not available for the control. Based on universal approximation results for radial basis function neural networks (RBF-NN), it has been proposed as an alternative to NN for approximating arbitrary nonlinear functions in L²(R). Adaptive RBF neural network is used to design a compensator for unmodeled dynamics in robotic system. Then asymptotically stability of the system is assured by combining nominal feedback controller and adaptive law of NN. The simulation results show the validity of the control scheme.