State estimation of distributed drive vehicles based on an improved adaptive extended Kalman filter

Vehicle state estimation is crucial in the development of high-order assisted driving systems as well as autonomous driving technologies. However, one of the difficulties it encounters is the cost–benefit ratio, or the idea of utilising less money to get the highest estimation outcomes. This study examines the problem of estimating the state parameter of distributed drive vehicles that are equipped with conventional sensors operating under high-speed driving conditions. It proposes an improved adaptive extended Kalman filter (IAEKF) method for state estimation that considers the impact of dynamic conditions on the noise covariance matrices Qand R. Additionally, it introduces a sliding window length that adapts Q and R, thereby mitigating the influence of non-smooth data on the estimation. By retaining measurements and estimation values within a specific time window, the algorithm enhances its practicality. To mitigate the differential error associated with the computation of the tyre longitudinal force, an adaptive sliding mode observer (ASMO) is employed to estimate the longitudinal force. The mean absolute error (MAE) and root mean square error (RMSE) of the sideslip angle and longitudinal velocity are analysed through simulation experiments by comparing the IAEKF to the EKF. The results demonstrate that the proposed method is superior.