第4期 宋锐，等：基于LiDAR/INS的野外移动机器人组合导航方法 ·809· 于机器人在进入小范围空间时，LOAM和LeGO SLAM based on 3D Lidar and RTK fusion[J].Manufactur LOAM的精度迅速下降，计算不同算法定位结果 ing automation,2020,42(7):51-54. 的均方根误差如表1所示，进一步验证了本文方 [5]CADENA C,CARLONE L,CARRILLO H,et al.Past, 法的有效性。 present,and future of simultaneous localization and map- ping:toward the robust-perception age[J].IEEE transac- 表1基于不同算法的实验结果均方根误差比较 tions on robotics,2016,32(6):1309-1332 Table 1 RMSE of the experimental results by different al- gorithms [6]YE H,CHEN Y,LIU M.Tightly coupled 3D lidar inertial odometry and mapping[C]//Proceedings of 2019 IEEE In- 算法 LeGO-LOAM LOAM scan-to-map ternational Conference on Robotics and Automation. X方向m 32.44 5.76 1.65 Montreal,Canada,2019:3144-3150. Y方向/m 25.54 5.04 1.03 [7]ZHANG J,SINGH S.Laser-visual-inertial odometry and mapping with high robustness and low drift[J].Journal of field robotics,2018,35(8):1242-1264 4总结与展望 [8]SHAN T,ENGLOT B.LeGO-LOAM:lightweight and 针对野外移动机器人环境建模和定位问题， ground-optimized lidar odometry and mapping on variable 本文从数据预处理、点云分割与特征提取、激光 terrain[Cl//Proceedings of 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems.Madrid. 里程计和信息融合等方法对组合系统进行分析， Spain,2018:4758-4765 结合移动机器人运动特性分析激光点云畸变等问 [9]HESS W,KOHLER D,RAPP H,et al.Real-time loop 题，提出基于IMU的预积分与scan-to-map的融合 closure in 2D LIDAR SLAM[C]//Proceedings of 2016 方法，通过构建优化函数并对每帧激光的位姿、 IEEE International Conference on Robotics and Automa- 优化初值进行更新，实验结果表明本文方法相对 tion.Stockholm,Sweden,2016:1271-1278. 于LeGO-LOAM、LOAM具有更好的环境建模和 [10]FORSTER C.CARLONE L,DELLAERT F,et al.On- 位置估计精度。此外，高精度的定位和地图构建 manifold preintegration for real-time visual-inertial odo- 对于运载体的智能行为有重要影响，地图中可能 metry[J].IEEE transactions on robotics,2017,33(1): 存在的移动物体会将建图问题进一步复杂化，下 1-21 一步考虑融合环境的语义信息，根据点云语义分 [11]BOSSE M.ZLOT R.FLICK P.Zebedee:design of a spring-mounted 3-D range sensor with application to mo- 割识别动态物体，将带有语义标签的物体进行数 bile mapping[J].IEEE transactions on robotics,2012, 据关联，利用几何信息构建约束关系，进而提高 28(5):11041119. 机器人的建图和定位精度。 [12]李帅鑫，李广云，王力，等.Lidar/IMU紧耦合的实时定 参考文献： 位方法[J/OL].自动化学报，https:/doi.org/10 16383/1.aas.c190424. [1]HIDALGO-CARRI6 J.POULAKIS P.KIRCHNER F.Ad- LI Shuaixin,LI Guangyun,WANG Li,et al.Lidar/IMU aptive localization and mapping with application to planet- tightly coupled real-time localization method[J/OL].Acta ary rovers[J].Journal of field robotics,2018,35(6): automatica sinica,https://doi.org/10.16383/j.aas.c190424. 961-987. [13]CHEN C.HE Y.GU F.et al.A Real-time relative probab- [2]YOUNG S H.MAZZUCHI T A.SARKANI S.A frame- ilistic mapping algorithm for high-speed off-road work for predicting future system performance in autonom- autonomous driving[C]//Proceedings of 2015 IEEE/RSJ ous unmanned ground vehicles[J].IEEE transactions on International Conference on Intelligent Robots and Sys- systems,man,and cybernetics:systems,2017,47(7): tems.Hamburg,Germany,2015:6252-6258. 1192-1206. [14]GENTIL CL,VIDAL-CALLEJA T,HUANG S.3D lid- [3]陈成，何玉庆，卜春光，等.基于四阶贝塞尔曲线的无人 ar-IMU calibration based on upsampled preintegrated 车可行轨迹规划[.自动化学报，2015,41(3)：486-496. measurements for motion distortion correction[Cl//Pro- CHEN Cheng,HE Yuging,BU Chunguang,et al.Feasible ceedings of 2018 IEEE International Conference on Ro- trajectory generation for autonomous vehicles based on botics and Automation.Brisbane,Australia,2018: quartic bezier curve[J].acta automatica sinica,2015,41(3): 2149-2155. 486-496. [15]GAO H.ZHANG X.YUAN J,et al.A novel global local- [4]倪志康，厉茂海，林容，等.基于三维激光雷达与RTK融 ization approach based on structural unit encoding and 合的SLAM研究)制造业自动化，2020,42(7)：51-54 multiple hypothesis tracking[J].IEEE transactions on in- NI Zhikang,LI Maohai,LIN Rui,et al.Research on strumentation and measurement,2019,68(11):于机器人在进入小范围空间时，LOAM 和 LeGO￾LOAM 的精度迅速下降，计算不同算法定位结果 的均方根误差如表 1 所示，进一步验证了本文方 法的有效性。 表 1 基于不同算法的实验结果均方根误差比较 Table 1 RMSE of the experimental results by different al￾gorithms 算法 LeGO-LOAM LOAM scan-to-map X方向/m 32.44 5.76 1.65 Y方向/m 25.54 5.04 1.03 4 总结与展望 针对野外移动机器人环境建模和定位问题， 本文从数据预处理、点云分割与特征提取、激光 里程计和信息融合等方法对组合系统进行分析， 结合移动机器人运动特性分析激光点云畸变等问 题，提出基于 IMU 的预积分与 scan-to-map 的融合 方法，通过构建优化函数并对每帧激光的位姿、 优化初值进行更新，实验结果表明本文方法相对 于 LeGO-LOAM、LOAM 具有更好的环境建模和 位置估计精度。此外，高精度的定位和地图构建 对于运载体的智能行为有重要影响，地图中可能 存在的移动物体会将建图问题进一步复杂化，下 一步考虑融合环境的语义信息，根据点云语义分 割识别动态物体，将带有语义标签的物体进行数 据关联，利用几何信息构建约束关系，进而提高 机器人的建图和定位精度。 参考文献： HIDALGO-CARRIó J, POULAKIS P, KIRCHNER F. Ad￾aptive localization and mapping with application to planet￾ary rovers[J]. Journal of field robotics, 2018, 35(6): 961–987. [1] YOUNG S H, MAZZUCHI T A, SARKANI S. A frame￾work for predicting future system performance in autonom￾ous unmanned ground vehicles[J]. IEEE transactions on systems, man, and cybernetics: systems, 2017, 47(7): 1192–1206. [2] 陈成, 何玉庆, 卜春光, 等. 基于四阶贝塞尔曲线的无人 车可行轨迹规划 [J]. 自动化学报, 2015, 41(3): 486–496. CHEN Cheng, HE Yuqing, BU Chunguang, et al. Feasible trajectory generation for autonomous vehicles based on quartic bezier curve[J]. acta automatica sinica, 2015, 41(3): 486–496. [3] 倪志康, 厉茂海, 林睿, 等. 基于三维激光雷达与 RTK 融 合的 SLAM 研究 [J]. 制造业自动化, 2020, 42(7): 51–54. NI Zhikang, LI Maohai, LIN Rui, et al. Research on [4] SLAM based on 3D Lidar and RTK fusion[J]. Manufactur￾ing automation, 2020, 42(7): 51–54. CADENA C, CARLONE L, CARRILLO H, et al. Past, present, and future of simultaneous localization and map￾ping: toward the robust-perception age[J]. IEEE transac￾tions on robotics, 2016, 32(6): 1309–1332. [5] YE H, CHEN Y, LIU M. Tightly coupled 3D lidar inertial odometry and mapping[C]//Proceedings of 2019 IEEE In￾ternational Conference on Robotics and Automation. Montreal, Canada, 2019: 3144−3150. [6] ZHANG J, SINGH S. Laser-visual-inertial odometry and mapping with high robustness and low drift[J]. Journal of field robotics, 2018, 35(8): 1242–1264. [7] SHAN T, ENGLOT B. LeGO-LOAM: lightweight and ground-optimized lidar odometry and mapping on variable terrain[C]//Proceedings of 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems. Madrid, Spain, 2018: 4758−4765. [8] HESS W, KOHLER D, RAPP H, et al. Real-time loop closure in 2D LIDAR SLAM[C]//Proceedings of 2016 IEEE International Conference on Robotics and Automa￾tion. Stockholm, Sweden, 2016: 1271−1278. [9] FORSTER C, CARLONE L, DELLAERT F, et al. On￾manifold preintegration for real-time visual-inertial odo￾metry[J]. IEEE transactions on robotics, 2017, 33(1): 1–21. [10] BOSSE M, ZLOT R, FLICK P. Zebedee: design of a spring-mounted 3-D range sensor with application to mo￾bile mapping[J]. IEEE transactions on robotics, 2012, 28(5): 1104–1119. [11] 李帅鑫, 李广云, 王力, 等. Lidar/IMU 紧耦合的实时定 位方法 [J/OL]. 自动化学报, https://doi.org/10. 16383/j.aas.c190424. LI Shuaixin, LI Guangyun, WANG Li, et al. Lidar/IMU tightly coupled real-time localization method[J/OL]. Acta automatica sinica, https://doi.org/10.16383/j.aas.c190424. [12] CHEN C, HE Y, GU F, et al. A Real-time relative probab￾ilistic mapping algorithm for high-speed off-road autonomous driving[C]//Proceedings of 2015 IEEE/RSJ International Conference on Intelligent Robots and Sys￾tems. Hamburg, Germany, 2015: 6252−6258. [13] GENTIL C L, VIDAL-CALLEJA T, HUANG S. 3D lid￾ar-IMU calibration based on upsampled preintegrated measurements for motion distortion correction[C]//Pro￾ceedings of 2018 IEEE International Conference on Ro￾botics and Automation. Brisbane, Australia, 2018: 2149−2155. [14] GAO H, ZHANG X, YUAN J, et al. A novel global local￾ization approach based on structural unit encoding and multiple hypothesis tracking[J]. IEEE transactions on in￾strumentation and measurement, 2019, 68(11): [15] 第 4 期 宋锐，等：基于 LiDAR/INS 的野外移动机器人组合导航方法 ·809·