Abstract: To ensure vehicle adaptive safety control, it is necessary to accurately estimate the vehicle driving state information. For some parameters,it is difficult to measure directly for both technical and economic reasons. This paper proposes an information fusion estimation strategy using low-cost sensors to estimate vehicle driving state. A 3 DOF nonlinear vehicle dynamics model including yaw, lateral and longitudinal directions is established. In order to reduce the impact of noise on the system, an adaptive unscented Kalman filter (AUKF) information fusion algorithm is established, and the fusion result in the sense of minimum variance of vehicle state is given. Low-cost sensors signal such as longitudinal acceleration, lateral acceleration and steering wheel angle is fused to obtain the required hard-to-measure sideslip angle of mass center, yaw angular velocity and longitudinal vehicle speed. The estimation algorithm is validated by the joint simulation of Matlab/Simulink-CarSim and the real vehicle test. The test results show that the algorithm can accurately estimate the sideslip angle, yaw angular velocity and longitudinal vehicle speed. Compared with UKF, the algorithm improves the estimation accuracy and real-time performance.

Key words: vehicle dynamics, state estimation, information fusion, AUKF