A novel vehicle dynamics identification method utilizing MIMU sensors based on support vector machine

Abstract

The major challenge of inertial navigation system (INS) is the rapid navigation error drift when aiding sensors are unavailable. However, if the dynamics of land vehicle can be detected, these errors can be corrected or restrained. A method based on support vector machine (SVM) using the outputs of MIMU is proposed here to identify the dynamics of land vehicle. This method computes part of the time-domain features and frequency-domain features. Then, a subset of these features is selected based on wrapper evaluation criteria. Afterwards, SVM is trained based on these selected features. Finally, the trained SVM is used in identification tests. The identification results show that this method can correctly identify the stationary, straight-line and cornering states.