AUKF based unified estimation scheme for non-linear vehicle dynamics

The paper presents an additive unscented Kalman filter (AUKF) based modular approach to estimate lateral vehicle dynamics and tyre forces. In this approach, a simplified single-track vehicular model and a dynamics-oriented tyre model are adopted to represent the vehicular motion. Subsequently, multiple observer modules for each dynamical state are designed and integrated into a unified estimation scheme (UES-AUKF). Two additional non-modular observers using AUKF and extended Kalman filter (EKF) are designed for comparative analysis of estimation accuracy and computational efficiency of the designed scheme. The simplified model and the designed estimators are simulated using double lane change (DLC) and sinusoidal (Sine) manoeuvres for high and low mu surfaces, respectively, and the results are analysed. Thereby the scheme is further validated using real vehicle dataset for estimation accuracy. Simulation results for the simplified vehicle and the tyre model conform with the standard results with acceptable deviations. Also, the proposed scheme exhibits improved accuracy with reduced computational time as compared to the non-modular observers.

Automated summary

This paper presents a modular approach based on AUKF to estimate lateral vehicle dynamics and tyre forces. Through simulation and real vehicle dataset validation, the results show that the proposed scheme has higher accuracy and less computational time compared to non-modular observers.