| 陈强,陈海波,张沥化.基于NDT配准与轮式里程计的激光雷达运动畸变补偿算法[J].高技术通讯(中文),2024,34(1):83~91 |
| 基于NDT配准与轮式里程计的激光雷达运动畸变补偿算法 |
| Lidar motion distortion compensation algorithm based on NDT registration and wheel odometry |
| |
| DOI：10. 3772/ j. issn. 1002-0470. 2024. 01. 009 |
| 中文关键词: 同时定位与地图构建（SLAM）； 运动畸变补偿； 激光雷达； 轮式里程计； 正态分布变换（NDT） |
| 英文关键词: simultaneous localization and mapping (SLAM), motion distortion compensation, lidar, wheel odometry, normal distributions transform (NDT) |
| 基金项目: |
| 作者 | 单位 | | 陈强 | （浙江工业大学信息工程学院杭州 310023） | | 陈海波 | | | 张沥化 | |
|
| 摘要点击次数: 964 |
| 全文下载次数: 802 |
| 中文摘要: |
| 激光雷达是广泛应用于同时定位与地图构建（SLAM）的测距传感器，普遍基于旋转机制收集周围环境的几何信息。当扫描期间激光雷达发生移动时，生成的点云会产生运动畸变，降低SLAM系统的准确性。在激光雷达SLAM算法设计中，为使雷达运动的估计结果更为精确，文中提出一种基于正态分布变换（NDT）和轮式里程计的激光雷达运动畸变补偿算法。首先，使用轮式里程计以高频测量方式对雷达运动进行估计，可补偿部分运动畸变。其次，设计一种基于NDT配准算法的误差处理方法，通过对点云的精准匹配降低里程计漂移的影响，实现雷达运动精确估计，进而精准补偿运动畸变。文中采用数据集以及真实场景实验对提出算法进行测试。实验结果表明，与传统里程计辅助方法相比，提出的算法能够优化运动畸变补偿效果，降低轨迹累积误差并生成全局一致地图。 |
| 英文摘要: |
| Lidar is a ranging sensor widely used in simultaneous localization and mapping (SLAM), and generally collects geometric information of the surrounding environment based on the rotation mechanism. When the lidar moves during the scan, the resulting point cloud will have motion distortions,which in turn reduces the accuracy of the SLAM system. In the SLAM algorithm of lidar, in order to make the estimation result of lidar motion more accurate, this paper proposes a motion distortion compensation algorithm of lidar based on normal distributions transform (NDT) and wheel odometry. Firstly, the lidar motion is estimated by high-frequency measurements using wheel odometry, which compensates for some motion distortions. Then, an error processing method is designed based on the NDT registration algorithm. By reducing the influence of odometry drift through accurate matching of point clouds, accurate motion estimation of lidar and accurate compensation for motion distortion can be achieved. The proposed algorithm is tested by datasets and real scene experiments. The experimental results show that, compared with the traditional odometry-assisted method, the proposed algorithm can optimize the motion distortion compensation effect, reduce the cumulative error of the trajectory and generate a globally consistent map. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
