激光雷达/MEMS IMU/里程计紧组合导航算法  

Lidar/MEMS IMU/Odometer integrated tightly navigation algorithm

作　　者：张福斌[1] 王凯 廖伟飞 孙成浩 Zhang Fubin;Wang Kai;Liao Weifei;Sun Chenghao(School of Marine Science and Technology,Northestern Polytechnical University,Xi'an 710000,China;Chinese Flight Test Establishment,Xi'an 710000,China)

机构地区：[1]西北工业大学航海学院,西安710000 [2]中国飞行试验研究院,西安710000

出　　处：《仪器仪表学报》2022年第7期139-148,共10页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金(51979228)项目资助。

摘　　要：为提高机器人在未知复杂环境中导航系统的鲁棒性与稳定性,提出了一种激光雷达/MEMS IMU/里程计紧组合导航算法。首先通过MEMS IMU/里程计的预积分,对激光雷达运动产生的畸变点云进行矫正,提高两帧点云之间的特征匹配效率;然后根据时间戳对预积分的机器人位姿进行线性插值,得到两帧点云之间粗略的位姿变化量,以此粗略的位姿变化量作为优化算法迭代初值,减少优化算法的迭代次数;其次在后端优化中加入MEMS IMU/里程计的运动约束,利用多传感器联合优化来提高机器人的定位精度;最后利用数据集进行仿真实验、利用四轮小车开展了室内与室外开闭环实验,实验表明,本算法室外开环定位误差均值比传统算法ALOAM、LEGO-LOAM分别减小51.01%和24.75%,并且其在拐弯等运动剧烈时能够保持较高精度。To improve the robustness and stability of the robot navigation system in an unknown and complex environment, a Lidar/MEMS IMU/Odometer integrated tightly navigation algorithm is proposed. Firstly, the algorithm corrects the distortion point cloud generated by the lidar movement through the pre-integration of the MEMS IMU/Odometer to improve the feature matching efficiency between two frames of the point cloud. Secondly, the linearly interpolation of the pre-integrated robot posture is implemented according to the timestamp to obtain the rough position change between two frames of the point cloud. This rough pose changing is used as the initial value of the optimization algorithm iteration to reduce the number of iterations of the optimization algorithm. Then, the motion constraint of MEMS IMU/Odometer is added to the back-end optimization, and the multi-sensor joint optimization is used to improve the positioning accuracy of the robot. Finally, the simulation experiment is carried out using the data set. The indoor and outdoor opening and closing loop experiments are implemented by using the four-wheeled trolley. Experiments show that the average outdoor open-loop positioning error of this algorithm is reduced by 51.01% and 24.75% respectively compared with the traditional algorithms ALOAM and LEGO-LOAM, respectively, and it can maintain high accuracy when the movement such as cornering is intense.

关 键 词：激光SLAM 紧组合 多传感器 点云畸变矫正 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置] TH76[自动化与计算机技术—控制科学与工程]