基于多层激光雷达的道路信息提取算法  

Environmental Perception of Multi-layer Laser Radar in Driverless Car

作　　者：段建民[1] 郑凯华[1] 李龙杰[1] 史丽晓 

机构地区：[1]北京工业大学智能测控研究所,北京100124

出　　处：《控制工程》2016年第4期468-473,共6页Control Engineering of China

基　　金：北京市自然科学基金资助项目(KZ20041000501)

摘　　要：为了获得无人驾驶车辆前方道路信息,应用多层激光雷达,提出一种道路信息提取算法。道路信息提取算法主要由路沿检测算法和道路坡度检测算法构成。路沿检测算法首先根据路沿数据点特征从众多的激光雷达数据中提取出路沿数据集,其次,应用基于欧氏距离改进的COBWEB算法对路沿数据集进行聚类分析,然后,提出多层融合规则将聚类后的路沿类分为左侧路沿和右侧路沿并剔除干扰路沿类,最后,应用最小二乘法拟合出左右路沿。道路坡度检测算法根据激光雷达不同扫描层上的路面数据点的相对位置关系,计算并得出无人驾驶车前方道路的坡度信息。实车测试表明了提出的道路信息提取算法可以稳定、准确的感知无人驾驶车前方道路信息。In order to obtain the road information ahead the driverless vehicle, an extraction algorithm is proposed with multi-layer laser radar. Road information extraction algorithm is mainly made up of road edge and slope detection algorithm. Firstly, the road edge data sets are extracted from numerous laser radar data based on characteristic of the road edge data points. Secondly, cluster analysis of the road edge data sets is done with the improved COBWEB algorithm based on Euclidean distance, and road edge classes are divided into left and right road edges according to multi-layer fusion rules. Finally, left and right road edges are fitted a straight line with the least squares method respectively. Slope detection algorithm is proposed to calculate the road slope according to the relative position of data points on different scanning layers. The results show that the extraction algorithm can stably and accurately perceive the road information around the driverless vehicle.

关 键 词：多层激光雷达 无人驾驶车 COBWEB算法 路沿检测 坡度检测 

分 类 号：TN958.98[电子电信—信号与信息处理]