Radial Basis Function Neural Network Sliding Mode Control with Improved Convergence Law for Robotic Manipulator

An adaptive neural network sliding mode control method based on improved reaching law is proposed to solve the unknown uncertainty and external interference of robotic manipulator control systems and the buffeting problem of traditional sliding mode control. Firstly, the mechanical arm is characterized through modeling, leading to the derivation of its dynamic equation. Subsequent to employing the sliding mode control strategy, an enhanced exponential reaching law is devised to expedite the approach of the system state to the sliding mode surface, consequently diminishing the response time of the system. Furthermore, leveraging the approximation characteristics of a radial basis function neural network (RBFNN), the modeling error and external interference are approximated. The inherent self-learning ability of the system overcomes uncertainties, consequently reducing the chattering phenomenon associated with sliding mode control. The efficacy of this approach is validated through simulation tests conducted on a two-link manipulator system.