This paper presents a vehicle sideslip angle estimation scheme against noises and outliers in sensor measurements for a four-wheel-independent-drive electric vehicle. The proposed scheme combines a robust unscented Kalman filter estimator based on the 3-DOF vehicle dynamics model and an extended Kalman filter estimator based on the kinematic model to form a hybrid estimator through a weighting factor. The weighting factor can be dynamically adjusted in real time to optimize the overall estimation performance under different driving conditions. The main contributions of this study to the related literature lie in two aspects. Firstly, a robust unscented Kalman filter estimator was incorporated to improve the robustness of dynamics-based estimation to sensor measurement outliers. Secondly, a novel moving polynomial Kalman smoother was included to filter out the noises in sensor measurements. Co-simulations of Matlab/Simulink and Carsim software were conducted under typical vehicle maneuvers and show that the proposed vehicle sideslip angle estimation scheme can obtain satisfied estimation results, with the moving polynomial Kalman smoother exhibiting better phase characteristics and filtering performance relative to commonly-used finite impulse response filters, and the robust unscented Kalman filter estimator being robust to sensor measurement outliers.
An Integrated Deep Ensemble-Unscented Kalman Filter for Sideslip Angle Estimation with Sensor Filtering Network