Abstract
This paper proposes a lane following control system that is robust to vehicle parameters variation, measurement noise and loss of sensor data. The loss of data is detected by the Chi-Square Test applied to the residual of an Auto-Regressive Model. The robustness of the control system is characterized by the probability of instability and probabilities of violation of design requirements. The structure of the control system is determined based on the Generalized Predictive Control theory. The design parameters of the controller are optimized by using Genetic Algorithm. Simulation results show that the optimized controller has good control performance and robustness.
