Vehicle Target Tracking Method for Crooked Roads Based on UKF and Sensor Fusion

Abstract

Aiming at the problem that mm-wave radar tracking target is easy to lose when in crooked roads, a vehicle target tracking method based on Unscented Kalman Filter (UKF) and sensor fusion is proposed. This method uses mm-wave radar and camera to obtain vehicle targets’ information in crooked roads at the same time. Then, according to whether mm-wave radar tracking target is lost or not, different tracking methods based on UKF are adopted. When the mm-wave radar tracking target is not lost, the historical and current data information of mm-wave radar are taken as the state information and measurement information, directly calculating the tracking vehicle parameters such as position, distance and speed; When the mm-wave radar tracking target is lost, a sensor fusion strategy is adopted. The historical data information of mm-wave radar is still the state information, and the camera vision image information of the tracking target becomes the measurement information, integrating the two sensors’ information to get the tracking vehicle parameters. And the result will serve as next moment’s state information for continuous tracking to solve the mm-wave radar tracking target loss problem in crooked roads. Finally, the validity of this method is verified by real road experiments.