摘要
为了满足室外移动机器人高速安全行驶的要求,提出了一种基于分段直线模型和增强型状态转移网络(ATN)的车道识别方法。给出了高速公路车道线的分段直线模型;在车道识别阶段,首先采用动态双阈值二值化方法对原始图像进行预处理得到二值化图像,再利用基于ATN的图像理解算法识别出车道线。使用该算法的清华大学室外移动机器人(THM R-V)系统在高速公路上的最高自主行驶速度已经超过了150 km/h,另外在各种天气和不同路况下的稳定表现也说明了该算法的鲁棒性。
A high speed lane detection method that satisfies the safety requirements of outdoor mobile robots was developed based on the segment-line road model and augmented transition networks （ATN）. After using a segment-line model to describe the lane marks, the lane detection phase uses a dynamic bi-threshold algorithm to get a binary image from original picture. The ATN-based image analysis method is then used to recognize the lane lines. The TsingHua mobile robot-V （THMR-V） system using this method has achieved a maximum automatic driving speed of 150 km/h on highways. Stable performance in various types of weather and road conditions has shown the robustnes of the system.
出处
《清华大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2006年第10期1762-1766,共5页
Journal of Tsinghua University(Science and Technology)
关键词
移动机器人
车道线检测
分段直线模型
ATN
mobile robot
lane detection
segment line model
augmented transition networks （ATN）
