A Loop Closure Detection Method Based on Point Cloud Semantic Graph Descriptor
With the continuous progress of computer level and robot sensor technology,mobile robots have gradually developed to intelligence.Among them,localization and mapping(SLAM)is the key technology of localization,and loop detection is an important module of SLAM system.In order to solve the problem of long time and low efficiency of traditional loopback detection,a loopback detection method based on point cloud semantic graph descriptor is pro-posed in this paper.The loopback detection performance is tested in the sequence of KITTI odometer data set 00 and 05,and compared with Scan Context,LEGO-LOAM and VFH algorithms.The results show that the method has the highest recall rate under 100％accuracy and 50％recall rate,and the best loopback detection performance.
point cloud descriptorsemantic segmentationloop closure detectionlaser SLAM
万方数据
维普
