可变高度激光里程计在室内不平整地面环境下的地图构建

针对可变高度激光雷达的结构特点提出了一种基于线、面特征的匹配算法, 并应用于存在不平整地面的结构化的室内环境. 首先提取激光点云中的边缘点和平面点, 并使用随机抽样一致性算法拟合成线段和平面; 然后根据线段与平面几何关系提出特征匹配条件, 完成特征之间的匹配; 利用特征间的匹配关系构建约束方程求解机器人位姿; 最后把机器人竖直方向位姿变换信息补偿到激光点云中, 并把纠正后的有一定高度的激光点云压缩成二维激光构建室内栅格地图. 实验结果表明, 与常用的二维激光同时定位与构图算法Gmapping对比, 在地面不平整的室内环境构建的地图得到了改善, 提高了移动机器人的构图精度.

Map construction of variable-height lidar odometry in indoor uneven ground environment

This paper proposes a matching algorithm on line and surface features by exploiting the characteristics of a variable-height lidar, which can be used in the structured indoor environment with uneven ground. Firstly, with the new algorithm, the edge points and plane points from the point cloud are obtained to fit line segments and planes. Based on the line segment and plane geometric constraints, feature matching conditions are obtained to build a constraint equation. The robot pose can then be calculated from the constraint equation. Finally, the calculated vertical pose transformation information of the robot is compensated, and used in the laser point cloud. The indoor grid map is constructed by a two-dimensional laser with a compressed laser point cloud. The algorithm is validated experimentally with Gmapping. A comparison with the variable-height lidar odometry algorithm is illustrated, which shows that the composition accuracy of the mobile robot is improved.