Vehicle state estimation based on Minimum Model Error criterion combining with Extended Kalman Filter

This paper researched an estimation method based on the Minimum Model Error (MME) criterion combing with the Extended Kalman Filter (EKF) for 4WD vehicle states. A general 5-input-3-output and 3 states estimation system was established, considering both the arbitrary nonlinear model error and the white Gauss measurement noise. Aiming at eliminating the estimation error caused by the arbitrary nonlinear model error, the prediction algorithm for the dynamic tire force error was deduced based on the MME criterion, based on which the system model can be effectively updated for higher estimation accuracy. The estimation algorithm was applied to a two-motor-driven vehicle during a double-lane-change process with varying speed under simulative experimental condition. The results showed that the dynamic tire force error could be effectively found for updating the system model, and higher estimation accuracy of the vehicle states were achieved, when compared with the traditional EKF estimator.