摘要
提出一种基于激光雷达点云的单帧单线滤波方法,基于滑窗机制的局部区域面片相似度分割方法,解决局部路面实时分割问题;提出一种单帧点云和局部点云的障碍物检测方法,解决机器人行驶中的实时障碍物检测问题;最后利用多个园区多个不同地形的场景进行测试,算法识别的准确率在95%以上,验证了本文方法的有效性和可靠性,具有重要的实际应用价值。
Aiming at the correct driving and intelligent obstacle avoidance problems of intelligent operation robots on unstructured roads such as local parks,we proposed a single-frame single-line filtering method based on LiDAR point cloud,and a local area patch similarity segmentation method based on sliding window mechanism to solve the problem of real-time segmentation of local pavement.And then,we proposed a single-frame point cloud and local point cloud obstacle detection method to solve the real-time obstacle detection problem in robot driving.Finally,we used multiple parks and different terrain scenarios for testing.The result shows that the recognition accuracy of algorithm is more than 95%,which verifies the effectiveness and reliability of the proposed method,and has important practical application value.
作者
郭永春
刘文博
罗作煌
GUO Yongchun;LIU Wenbo;LUO Zuohuang(Shenzhen Yijiahe Technology R&D Co.,Ltd.,Shenzhen 518055,China)
出处
《地理空间信息》
2022年第12期23-26,共4页
Geospatial Information
关键词
局部感知
激光点云
可行驶区域
滑窗法
local perception
laser point cloud
drivable area
sliding window
