Vehicle Tracking and Motion Estimation on Curve Road Segment by Using Smartphone Sensors

Abstract

This paper resolves the modelling problem of target vehicle's lateral motion in real-traffic tracking applications. We incorporate CTR (Constant turn rate) model with UKF (Unscented Kalman Filter) algorithm to deal with the vehicle's turning maneuver on curve road segment for a high accurate vehicle motion estimation. In order to test the proposed model's accuracy and applicability under complex real road environment, we design a traffic simulation experiment and drive the test vehicle to pass a turning road. The yaw rate and position information, which could be collected by smartphone sensor (gyroscopes, GPS and orientation sensor), are considered as observed value. Then, vehicle motion trajectory can be accurately obtained after the computation of the proposed model. Comparing with other vehicle motion information fusion algorithms, the result shows the proposed model presents a better accuracy in estimating and predicting vehicle lateral motion state on curve road segment.