点云空间与反射强度融合的非结构化道路可行驶区域检测

Detection of drivable areas on unstructured roads fused with point cloud space and reflection intensity

中文关键词  智能驾驶  非结构化道路  三维激光雷达  反射强度  空间分布  区域检测

英文关键词  intelligent driving  unstructured road  3D-LIDAR  reflection  spatial distribution  area detection

中文摘要

      可行驶区域检测旨在检测和提取智能车辆在道路上可行进的区域, 目前主流的检测方法主要基于三维激光雷达的空间特征实现, 难以处理路面边缘无清晰空间特征的非结构化道路. 为此, 本文提出了一种基于点云空间和反射强度融合的非结构化道路可行驶区域检测方法. 首先, 通过融合反射强度因子改进了基于空间特征的柱坐标系检测模型; 然后, 使用强度和降维空间检测对检测精度较低的环形检测模型进行优化, 并将其与柱坐标系检测模型联合使用以提高方法检测准确率; 最后, 在自录实际道路数据集上进行对比实验. 实验结果表明本文方法显著提高了非结构化道路可行驶区域检测的成功率与精确率, 在结构化道路上也具有良好效果.

英文摘要

      Drivable area detection aims to detect and extract areas where intelligent vehicles can travel on the road. The current mainstream detection method is mainly based on the spatial feature of three-dimensional light detection and ranging (3D-LIDAR), which is difficult to deal with unstructured roads without clear spatial features at the edge of the road surface. To this end, this paper proposes a drivable area detection method for unstructured roads based on the fusion of point cloud space and reflection intensity. First, the cylindrical coordinate system detection model based on the spatial features is improved by fusing reflection intensity factors; then, using intensity and dimensionality reduction space detection to optimize the ring detection model with low detection accuracy, and combining it with the cylindrical coordinate system detection model to improve the detection accuracy of the method; finally, a comparative experiment is carried out on the self-recorded actual road dataset. The experimental results show that the method in this paper significantly improves the success rate and accuracy of the drivable area detection on unstructured roads, and it also has good results on structured roads