Adaptive Parameter Approaching Law-Based Sliding Mode Control for Wheeled Robots

摘要

A method combining adaptive parameter approaching law with Radial Basis Function (RBF) neural network hierarchical sliding mode control is proposed to address the issues of low accuracy and significant oscillations during the trajectory tracking process of wheeled inverted pendulum robots (WIPR) with nonlinear underactuation characteristics. Compared to the traditional approach of combining approaching law with neural network sliding mode control, the innovation lies in the adaptive adjustment of the velocity of the motion point based on its distance to the sliding mode surface. By integrating the advantages of hierarchical sliding mode control and neural networks, the method effectively tracks the target trajectory. For the control system of multivariable complex robots, it is decomposed into second-order underactuated subsystems and first-order fully actuated subsystems, and the stability of the designed system is verified using Lyapunov functions. Simulations and comparative experiments are conducted using Matlab/Simulink. The results indicate that, compared to the method combining exponential approaching law, saturation approaching law, and RBF neural network hierarchical sliding mode control, this method exhibits better tracking performance and effectively suppresses oscillations generated during the motion process, demonstrating improved stability when the system is subjected to disturbances.