基于改进Cartographer的激光SLAM算法

Laser SLAM algorithm based on improved Cartographer

针对Cartographer算法在激光雷达的数据处理中存在的点云特征丢失的问题和低帧率激光雷达导致的运动畸变问题,提出一种改进激光同步定位与地图构建(simultaneous localization and mapping, SLAM)算法。采用k邻域搜索邻近点降采样的体素滤波方法代替Cartographer算法中的传统体素滤波方法,在不丢失点云特征的情况下提升计算速率;嵌入轮式里程计辅助模块去除激光雷达运动畸变,减少机器人的位姿累积误差,从而改善建图效果;最后,增加了边约束条件改善回环检测效果。通过在机器人操作系统中的gazebo搭建仿真环境进行模拟实验,对比两种算法,实验结果显示改进算法的建图轨迹误差更小。

An improved laser simultaneous localization and mapping(SLAM)algorithm was proposed to address the problem of point cloud feature loss and motion distortion caused by low frame rate lidar in Cartographer algorithm.The traditional voxel filtering method in Cartographer algorithm was replaced by k-neighborhood search for neighboring points and down-sampling,which improved the computational rate without losing point cloud features.The embedded wheeled odometer assist module removed lidar motion distortion and reduced the cumulative error of the robot′s pose,thus improving the map building effect.Finally,the edge constraint conditions were added to improve the loop closure detection effect.Simulations were carried out by building a simulation environment in gazebo in the robot operating system to compare the two algorithms.The experimental results show that the improved algorithm has less error in building the trajectory.