Unknown input observer for vehicle lateral dynamics based on a Takagi-Sugeno model with unmeasurable premise variables

Abstract : In this paper, an observer is proposed for estimating vehicle states which are not available for measurement. A new representation of a nonlinear vehicle model by exact Takagi-Sugeno structure is proposed based on sector nonlinearity approach. It leads to a multiple model with unmeasurable premise variables which constitutes the first contribution of the present paper. As a second contribution, a state estimation by an Unknown Input Observer (UIO) with unmeasurable premise variables is considered. It takes into account the variations of the longitudinal velocity in order to reflect realistic driving conditions. The convergence conditions of this observer are formulated as an optimization problem subjected to constraints established using Lyapunov theory and expressed under Linear Matrix Inequalities. Finally, some simulations are performed on the nonlinear model of the vehicle including the variations of the longitudinal velocity.