Adaptive Neural Network Control for a Robotic Manipulator with Unknown Deadzone

In this paper, adaptive neural network control is designed for a robotic manipulator with unknown dynamics. Neural networks are used to compensate for the unknown deadzone effect faced by the manipulator's actuator. State-feedback control is proposed first and high-gain observer is then designed to make the proposed control scheme more practical. The deadzone effect is approximated by a Radial Basis Function Neural Network (RBFNN) and the tracking error for the deadzone effect is bounded and converging. The unknown dynamics of the robotic manipulator is estimated with another RBFNN. Compensating for the estimated deadzone effect in the control law then leads to our proposed control. The proposed control is then verified on a two-joint rigid manipulator via numerical simulations.