A tube-based model predictive control method for intelligent vehicles path tracking.

The traditional Model Predictive Control (MPC) algorithm, grounded in precise mathematical models, faces challenges attributed to uncertainties in vehicle model parameters and modeling errors. These limitations result in suboptimal accuracy of the path tracking controller, particularly in complex driving environments. To address this issue, a tube-based model predictive control strategy (Tube MPC) is proposed. Initially, nominal dynamic modeling of intelligent vehicles is conducted, followed by linearized derivations. Leveraging the resulting linearized nominal model, the MPC cost function is solved to derive the control law for the nominal system. Subsequently, an error model between the actual and nominal systems is established, integrating various target constraints for path tracking and formulating a robust objective function. Utilizing the linear matrix inequality optimization method, a state feedback-assisted control law is derived to mitigate steady-state lateral position error. This approach ensures a more precise alignment between the actual and nominal states, ultimately enhancing tracking accuracy. Finally, experimental validations are conducted in the Carsim/Simulink simulation environment, encompassing diverse vehicle speed conditions and driving scenarios to assess the proposed controller against traditional Model Predictive Control. The experimental results demonstrate that, while ensuring vehicle stability, this control method exhibits superior tracking accuracy. [ABSTRACT FROM AUTHOR]