Abstract
In the advanced driver assistance system (ADAS), millimeter-wave radar is an important sensor to estimate the motion state of the target-vehicle. In this paper, the estimation of target-vehicle motion state includes two parts: the tracking of the target-vehicle and the identification of the target-vehicle motion state. In the unknown time-varying noise, non-linear target-vehicle tracking faces the problem of low precision. Based on the square-root cubature Kalman filter (SRCKF), the Sage–Husa noise statistic estimator and the fading memory exponential weighting method are combined to derive a time-varying noise statistic estimator for non-linear systems. A method of classifying the motion state of the target vehicle based on the time window is proposed by analyzing the transfer mechanism of the motion state of the target vehicle. The results of the vehicle test show that: (1) Compared with the Sage–Husa extended Kalman filtering (SH-EKF) and SRCKF algorithms, the maximum increase in filtering accuracy of longitudinal distance using the improved square-root cubature Kalman filter (ISRCKF) algorithm is 45.53% and 59.15%, respectively, and the maximum increase in filtering the accuracy of longitudinal speed using the ISRCKF algorithm is 23.53% and 29.09%, respectively. (2) The classification and recognition results of the target-vehicle motion state are consistent with the target-vehicle motion state.
Highlights
Millimeter-wave radar is an important sensor that constitutes an advanced driver assistance system (ADAS)
This paper studies the target tracking algorithm based on the millimeter-wave radar Cartesian coordinate system
This paper studies the target tracking algorithm based on the millimeter-wave coordinate system, to verify the performance and accuracy of the proposed algorithm in the target radar Cartesian coordinate system, to verify the performance and accuracy of the proposed tracking process
Summary
Millimeter-wave radar is an important sensor that constitutes an advanced driver assistance system (ADAS). Extended Kalman filter (EKF) [1,2], unscented Kalman filter (UKF) [3,4], particle filter (PF) [5] and cubature Kalman filter (CKF) [6] are common non-linear filtering state estimation algorithms. The motivation of writing the paper is as follows: (1) For the on-board millimeter-wave radar in the unknown and time-varying noise environment, the accuracy of a high-dimensional non-linear target tracking process is low. The ISRCKF algorithm based on SRCKF is proposed to accurately estimate the unknown and time-varying noise statistics.
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call