Approaches to Estimation of Vehicle Lateral Dynamics

In this paper, two new sliding mode state observers are developed for the lateral dynamics of road vehicles. It is shown that the observability of a general class of second-order linear time-varying systems (LTVSs) is investigated, the sliding observers for the LTVSs with both the known-input and the unknown-input are then proposed. The main contribution of this research is to use the learning concept from artificial intelligence for designing the sliding mode observers in the following ways: (i) The observer for the leading sub-system is defined with the finite time error convergence; (ii) The observers for other sub-systems are then independently designed to follow the desired error dynamics of the leading sub-system. Particularly, for the LTVSs with the unknown input, two extended states are added and the same learning strategy is adopted to construct the extended sliding observers, to ensure that both the system states and the unknown inputs can be accurately estimated in finite time. The simulations for the lateral dynamics of a typical road vehicle are conducted to confirm the advantages and effectiveness of the proposed sliding observer algorithms.