基于扩展卡尔曼滤波的车道融合跟踪

车道检测算法的研究是智能车辆自动导航的首要环节。与目前基于视觉的车道检测与跟踪系统不同,本文提出一种基于扩展卡尔曼滤波的车道融合跟踪方法。该方法利用毫米波雷达探测到前方车辆的距离信息,并采用扩展卡尔曼滤波技术和图像处理技术,建立车道跟踪的动态视觉窗口,提取车道边界,并判断前方车辆相对于车道的位置。该方法大大缩减了处理时间,且增强了系统的鲁棒性。

Lane Fusion Tracking System with Extended Kalman Filter

It's a principal link to detect lane robustly and rapidly under a wide variety of conditions for intelligent vehicle navigation,which posed a challenge for current lane detection and tracking systems focusing on vision-based algorithms.The paper proposed a lane fusion tracking method based on extended Kalman filter.The method first made use of the range information detected by millimeter radar,and estimated the obstacle's position of next frame relative to host vehicle with extended Kalman filter.Then,the projective relationship between image and the real world was used to build dynamic visual window for lane tracking,and finally lane boundary markings were extracted using vision technology,and the relative position of vehicle ahead to the lane was determined.As image processing was made only within the window,the time cost was significantly reduced which made it better satisfying realtime request,meanwhile it strengthened the system's robustness for shadowed,broken or interrupted lane.