中文摘要：

针对结构化道路环境中道路边界存在不连续、被遮挡及易受路内障碍物干扰情况下的识别问题,利用车 载激光雷达获取的结构化道路环境三维点云数据的高程信息,结合局部均值变点统计方法,提出了一种用于 激光雷达帧数据的道路边界识别算法.该算法首先利用局部均值变点统计对结构化道路环境三维点数据中 突变的z 坐标值进行标记并提取其对应的（ x, y ）数据点,即道路边界点数据粗提取;然后基于分段双阈值对 粗提取的道路边界点数据滤波处理;最后利用最小二乘法拟合道路边界点数据.基于实车实验分别采集的不 同道路环境条件下结构化直道1 4 5 0帧、弯道9 3 5帧数据,算法识别准确率均高于80%,且识别道路宽度误差 小于0. 14 m.实验结果表明,该算法不仅能够自动识别结构化道路边界,而且有效抑制了路面障碍物的干 扰,验证了算法的有效性.

英文摘要：

Aiming at the road boundary identification for discontinuous,occluded and susceptible to obstructions in the road structured road environment, a road boundary identification algorithm for LIDAR （ Light Deteation and Ranging） data frames is proposed by using the local mean change point statistics based on the elevation information of structured road environment points acquired by the onboard lidar. Firstly,the local mean change point statistics is used to mark the z-coordinate of the mutation in the data of the structured road environment and to extract the corresponding （x , y） data points,rough extraction of the road boundary point data. Then,the segmented dual threshold method is used to filter the road boundary points. Finally, the least squares method is used to fit the road boundary point data. Based on the laser radar acquisition of structured straight road 1450 frame data and bending road 935 data under different road environment conditions, the accuracy of the algorithm is higher than 80% , and the recognition error of the road width is less than 0.14 m . The experimental results show that the algorithm can not only automatically identify the structured road boundary, but also effectively restrain the obstacle interference, and verify the effectiveness of the algorithm.