Design of a vehicle steering assist controller using driver model uncertainty

Abstract

Single vehicle road departure (SVRD) accidents lead to over 1/3 of highway fatalities in the United States. A steering assist controller to reduce SVRD accidents is proposed. Previous studies using driving simulator data illustrate that the parametric uncertainty can represent the change in driver steering behaviour during long driving task. The uncertainty results show that occasionally the driver can perform very poorly. A serial steering control approach is investigated. The controller is used to attempt robust performance under the influence of driver uncertainty. A robust Smith predictor controller is employed to handle the delay in the driver model. The results indicate that the poor stability situations are successfully avoided and robust stability is achieved. However, due to large variations in driver behaviour, robust performance is not achieved. An adaptive version of the robust Smith predictor controller is shown to reduce the uncertainty in the driver model. The designed controllers are evaluated using frequency domain analyses and computer simulations.