基于DDP的激光回环检测算法

Laser loop closure detection based on DDP descriptor

回环检测作为SLAM技术重要组成部分,能够提高轨迹估计的全局一致性。针对目前效果较好的扫描上下文算法(SC),本文提出一种基于DDP改进的回环检测算法。首先,根据点云分布对齐点云坐标系;其次,利用方位角与尺度信息构建DDP点云描述符;最后,提出高度维度描述符余弦距离与离散度描述符相关距离加权方法,计算候选帧DDP点云描述符相似度。实验结果表明,本文所提出的基于DDP改进的回环检测算法在KITTI数据集大多数序列中,精确率与召回率优于IRIS、NDD、SC和ISC算法,且能够满足激光雷达实时性要求。

Loop closure detection,as an important part of SLAM technology,can improve the global consistency of trajectory estimation.In response to the issues existing in the currently effective Scan Context algorithm(SC),this paper proposes a loop closure detection algorithm based on DDP improvement.Firstly,align the point clouds coordinate system based on the point cloud distribution;secondly,construct DDP point cloud descriptors using azimuth and scale information;finally,propose a weighted method for cosine distance of height-dimensional descriptors and related distance of discreteness descriptor to calculate the similarity of candidate frame DDP point cloud descriptors.Experimental results show that the loop closure detection algorithm based on DDP improvement performs better in most sequences of the KITTI dataset,with higher precision and recall rates than IRIS,NDD,SC,and ISC algorithms,and can meet the real-time requirements of LiDAR.