Abstract
This brief presents a contribution to nonlinear observer design for both the estimation of vehicle lateral dynamics and road curvature. The latter is recovered by a simple algebraic technique and high-order sliding mode differentiator, which allows to estimate exactly the time derivatives of measured signals. For the lateral dynamics of the vehicle, a new approach is developed by the use of a Takagi-Sugeno (TS) model representing exactly, with no loss of information, the nonlinear vehicle behavior in a compact set of the state space. The TS model involves unmeasured premise variables. The proposed new observer starts with the estimation of these premise variables. Second, the design of an observer with weighting functions depending on these estimated premise variables is considered. Theoretically, the proposed observer ensures exponential convergence of the state estimation error toward zero. This convergence is studied by the Lyapunov theory and the obtained stability conditions are expressed in terms of linear matrix inequalities. Finally, experimental results are given for vehicle lateral dynamics and road curvature estimation with real data.