An improved lane departure warning algorithm based on fusion of F-Kalman filter and F-TLC

Abstract

In order to reduce the lane departure accident caused by driver’s negligence, LDWS (lane departure warning system) has become increasingly popular and important. However, most researches proposed mainly focus on how to detect lane markings. In this paper, we propose an improved lane departure warning algorithm based on fusion of F-Kalman filter (kalman filter based on fuzzy logic) and F-TLC (time to lane crossing based on fuzzy logic). First of all, least square method is used to calculate the distance between vehicle and lane markings. Then the estimation of vehicle states in the future is generated by means of the traditional kalman filter. To better work for lateral offset estimation, a fuzzy model is adopted to change the size of covariance matrices, which is used to adjust the traditional kalman filter in time. Finally, we further put forward to utilize F-TLC to generate multi-grade alarm. Extensive experiments are conducted on different conditions. Experimental results indicate that our warning method works efficiently. The average time consumed for system in each frame is 20.0216 ms. The proposed method possesses good robustness, and can be widely us\ed in LDWS.