A Feedforward-Feedback Coordinated Control-based Actuator Error Compensation Strategy

Autonomous vehicles require accurate and robust tracking of reference paths to ensure driving safety. To achieve better path tracking performance, improvements are needed not only in control algorithms but also in mitigating the adverse effects of actuator delays and various disturbances on the transmission of upper-level control commands. This paper presents a feedforward-feedback coordinated control-based actuator error compensation strategy.A delay neural network is employed to predict the time-delay effects of the steering system, and a feedforward compensator is designed to pre-compensate the inputs. Additionally, based on the current execution error, a Radial Basis Function Neural Network-Proportional-Integral-Derivative (RBFNN-PID) state feedback controller is developed to eliminate the impact of external disturbances that the feedforward compensator cannot predict. Experimental results demonstrate that the proposed coordinated control strategy effectively overcomes the effects of steering system delays and non-system factors such as external environmental influences, accurately generates the required front wheel angles, and significantly enhances the path tracking accuracy of autonomous vehicles.