随着智能机器人技术的快速发展，加快了测绘行业智能化、实时化的改革进程。其 中，未知领域环境下的机器人同步定位与地图构建技术（Simultaneous Localiz Ationand Mapping, SLAM）成为整个行业的热点研究方向，推动了传统测绘向着实时动态的方向 发展，但经典 SLAM 算法仅依靠视觉里程计进行位置状态估计，易受光照、环境纹理和 动态场景等环境因素的影响，无法维持长时效、远距离的高精度定位，容易出现误差累 积定位不准确等问题。因此，本文提出一种基于人工地标约束的方法，对处于未知环境 中的机器人进行同步定位与建图技术研究，主要研究成果如下:

     （1）针对运动约束系统存在的多个坐标系之间的转换问题，本文探讨了不同坐标 系下的坐标表达以及相互转换关系，实现了坐标关系的统一；对旋转矩阵存在的自身约 束问题，探讨了基于李群李代数的位姿表达数学形式，并进行机器人位姿的局部非线性 优化，得到局部最优解。 

    （2）为解决视觉传感器在 AprilTag 码检测识别过程中存在的误识别、漏识别等问 题，本文在图像预处理基础上，对图像进行改进滤波处理，提高图像边缘信息的保留度， 结合四边形优化算法和 Canny 算子提高图像边缘提取的准确性，通过识别解码实现人工 地标位置信息获取。针对实际应用场景中 AprilTag 码的不同形态，进行识别鲁棒性和精 度的验证实验。实验结果表明，与传统识别算法相比，本文算法识别鲁棒性和精准性更 高，识别成功率提高了 5.40%，为后续约束机器人定位提供保障。 

    （3）针对经典 SLAM 技术出现的误差累积问题，本文提出一种人工地标约束的移 动机器人 SLAM 定位方法。通过相机对场景中布设的 AprilTag 码进行精准识别，结合 视觉里程计估计的机器人运动状态，建立运动位姿约束模型，实现机器人运动状态的全 局优化，获得最优位姿，并进行本文定位方法和经典 SLAM 定位方法的精度对比实验。 实验结果表明，在直线和环形两种不同运动形式下，本文提出的约束定位方法较经典 SLAM 定位方法均方根误差分别降低了 5.51%和 38.23%，验证了本文方法的有效性。 

      With the rapid development of intelligent robot technology, the process of intelligent and real-time reform in the surveying and mapping industry has been accelerated. Among them, the Simultaneous Localiz Action and Mapping (SLAM) technology of robots in the unknown environment has become a hot research direction in the entire industry, which has promoted the development of traditional surveying and mapping in the direction of real-time dynamics. However, the classical SLAM algorithm only relies on visual odometry. The location state estimation is easily affected by environmental factors such as lighting, environmental textures and dynamic scenes. Therefore, this paper proposes a method based on artificial landmark constraints to conduct synchronous positioning and mapping technology research on robots in unknown environments. The main research results are as follows: 

      (1) For the conversion problem between multiple coordinate systems in the motion constraint system, this paper discusses the coordinate expression and mutual conversion relationship in different coordinate systems, and realizes the unification of the coordinate relationship; for the self-constraint problem of the rotation matrix, The mathematical form of pose expression based on Lie group Lie algebra is discussed, and the local nonlinear optimization of the robot pose is carried out to obtain the local optimal solution.     

      (2) In order to solve the problems of misrecognition and missed recognition of the vision sensor in the process of AprilTag code detection and recognition, based on image preprocessing, this paper improves the filtering of the image, improves the retention of image edge information, and combines quadrilateral optimization. The algorithm and Canny operator improve the accuracy of image edge extraction, and realize the acquisition of artificial landmark position information through recognition and decoding. According to different forms of AprilTag codes in practical application scenarios, verification experiments of recognition robustness and accuracy are carried out. The experimental results show that compared with the traditional recognition algorithm, the algorithm in this paper has higher recognition robustness and accuracy, and the recognition success rate is increased by 5.40%, which provides a guarantee for the subsequent constrained robot positioning. 

      (3) Aiming at the error accumulation problem of classical SLAM technology, this paper proposes a SLAM localization method for mobile robots constrained by artificial landmarks. The AprilTag code laid out in the scene is accurately identified by the camera, combined with the motion state of the robot estimated by the visual odometry, the motion pose constraint model is established, the global optimization of the motion state of the robot is realized, the optimal pose is obtained, and the positioning method and Accuracy comparison experiments of classic SLAM localization methods. The experimental results show that the constrained localization method proposed in this paper reduces the root mean square error by 5.51% and 38.23%, respectively, compared with the classical SLAM localization method under two different motion forms, namely linear and circular, which verifies the effectiveness of the proposed method.

[1] 国务院关于印发新一代人工智能发展规划的通知[J]. 中华人民共和国国务院公报,

2017(22): 7-21.

[2] 李德仁, 张洪云, 金文杰. 新基建时代地球空间信息学的使命[J/OL]. 武汉大学学报(信息科学版): 1-10[2022-04-16].

[3] 权美香, 朴松昊, 李国. 视觉 SLAM 综述[J]. 智能系统学报, 2016, 11(6): 768-776.

[4] 邸凯昌, 万文辉, 赵红颖, 等. 视觉 SLAM 技术的进展与应用[J]. 测绘学报, 2018,47(06): 770-779.

[5] 王延东. 多位姿信息融合的双目视觉惯性里程计研究[D]. 长春: 中国科学院大学(中国科学院长春光学精密机械与物理研究所), 2019.

[6] 朱枫. 基于卡尔曼滤波提高陀螺传感器测量精度的研究[D]. 长春: 中国科学院长春光学精密机械与物理研究所, 2017.

[7] Leonard J J,Durrant-Why H F, et al.. Simultaneous map building and localization for an autonomous mobile robot [M]. DEEE/RSJ International Workshop on Intelligent Robots and Systems , 1991: 1442-1447.

[8] Gutmann J S, Konolige K . Incremental mapping of large cyclic environments[C]. Proc the Conference on Intelligent Robots & Applications. 1999.

[9] Holmes S, Klein G, Murray D W. A Square Root Unscented Kalman Filter for visual mono SLAM[C]. IEEE International Conference on Robotics and Automation. IEEE, 2008:3710-3716.

[10]Aoki E H, Bagchi A, Mandal P, et al. A theoretical look at information-driven sensor management criteria[C]. 14th International Conference on Information Fusion. IEEE, 2011.

[11]Bardow P, Davison A J, Leutenegger S. Simultaneous Optical Flow and Intensity Estimation from an Event Camera[C]. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE Computer Society, 2016.

[12]Davison A. J. Real-time simultaneous localisation and mapping with a single camera[C]//9th IEEE International Conference on Nice. France: IEEE, 2003: 1403-1410.

[13]Mourikis A, Roumeliotis S. A multi-state constraint kalman filter for vision-aided inertial navigation[C]//Proceedings of IEEE International Conference on Robotics and Automation. Los Alamitos: IEEE Computer Society Press, 2007: 3565-3572.

[14]Lynen S, Achtelik M W, Weiss S, et al. A robust and modular multi-sensor fusion approach applied to MAV navigation[C]//IEEE/RSJ International Conference on Intelligent Robots and Systems. Piscataway, USA: IEEE, 2013: 3923-3929.

[15]Klein G, Murray D. Parallel tracking and mapping for small AR workspaces[C]//IEEE and ACM International Symposium on Mixed and Augmented Reality, 2007: 225-234.

[16]Mourikis A I, Roumeliotis S I. A Multi-State Constraint Kalman Filter for Visual-aided Inertial Navigation[C]// IEEE International Conference on Robotics and Automation. Italy: IEEE, 2007: 3565-3572.

[17]Forster C , Carlone L , Dellaert F , et al. On-Manifold Preintegration for Real-Time Visual-Inertial Odometry[J]. ieee transactions on robotics, 2017, 33(1): 1-21.

[18]Mur-Artal R, Montiel J, Tardos J. ORB-SLAM: A versatile and accurate monocular SLAM system[J]. IEEE Transactions on Robotics, 2015, 31(5): 1147-1163.

[19]C. Forster, M. Pizzoli, et al.SVO: Fast Semi-Direct Monocular Visual Odometry[J].IEEE International Conference on Robotics and Automation (ICRA), 2014, 23(4)15-22.

[20]Mur-Artal R, Tardós J D.Visual-inertial monocular SLAM with map reuse[J].IEEE Robotics & Automation Letters, 2017, 2(2): 796-803.

[21]Campos C , Elvira R , JJG Rodríguez, et al. ORB-SLAM3:An Accurate Open-Source Library for Visual, Visual-Inertial and Multi-Map SLAM[J]. 2020.

[22]林睿. 基于图像特征点的移动机器人立体视觉 SLAM 研究[D]. 哈尔滨: 哈尔滨工业大学. 2011.

[23]刘艳丽. 融合颜色和深度信息的三维同步定位与地图构建研究[D]. 长沙: 中南大学,2014.

[24]Zhou H ,Zou D ,Pei L , et al. StructSLAM: Visual SLAM With Building Structure Lines[J]. IEEE Transactions on Vehicular Technology, 2015, 64(4): 1364-1375.

[25]Qin T, Li P, Shen S. VINS-Mono: a robust and versatile monocular visual-inertial state estimator [J]. IEEE Transactions on Robotics, 2018, 34(4): 1004-1020.

[26]刘毅. 基于单目视觉的实时高精度定位方法研究[D]. 华中科技大学, 2019.

[27]傅柳军, 龚烨, 李礼, 等. 面向高动态环境的 ORB-SLAM3 算法优化[J]. 测绘地理信息, 2021, 46(S01): 93-96.

[28]邓晨, 李宏伟, 张斌, 等. 基于深度学习的语义 SLAM 关键帧图像处理[J]. 测绘学报, 2021, 50(11): 1605-1616.

[29]Klopschitz M , Schmalstieg D . Automatic Reconstruction of Wide-Area Fiducial Marker Models[C]// IEEE & Acm International Symposium on Mixed & Augmented Reality. ACM, 2007.

[30]任沁源. 基于视觉信息的微小型无人直升机地标识别与位姿估计研究[D]. 杭州: 浙江大学信息学院 浙江大学, 2008.

[31]Lim H , Lee Y S . Real-time single camera SLAM using fiducial markers[C] Iccas-sice. IEEE, 2009.

[32]温丰, 柴晓杰, 朱智平, 等. 基于新型人工路标系统的视觉 SLAM 算法研究[C]//. 第二十九届中国控制会议论文集, 2010: 3866-3872.

[33]Karam Shaya, Aaron Mavrinac, Jose Luis Alarcon Herrera. A Self-localization System with Global Error Reduction and Online Map-Building Capabilities[J]. 2012.

[34]Chiou Y S, Tsai F, Yeh S C, et al. An IMU-based positioning system using QR-code assisting for indoor navigation[M]. Computer Science and itsApplications. Springer, Dordrecht, 2012: 655-665.

[35]Lee S J , Lim J , Tewolde G , et al. Autonomous tour guide robot by using ultrasonic range sensors and QR code recognition in indoor environment[C] IEEE International Conference on Electro/information Technology. IEEE, 2014.

[36]Zhang Z , Meng X . Use of an Inertial/Magnetic Sensor Module for Pedestrian Tracking During Normal Walking[J]. IEEE Transactions on Instrumentation and Measurement, 2015, 64(3): 776-783.

[37]Nazemzadeh P , Fontanelli D , Macii D , et al. Indoor Localization of Mobile Robots through QR Code Detection and Dead Reckoning Data Fusion[J]. IEEE/ASME Transactions on Mechatronics, 2017: 1-1.

[38]王永力, 胡旭晓, 兰国清, 等. 基于视觉 SLAM 和人工标记码的定位与导航方法研

究[J]. 成组技术与生产现代化, 2018, 35(04): 22-26.

[39]Pfrommer B , Daniilidis K . TagSLAM: Robust SLAM with Fiducial Markers[J]. 2019.

[40]Shamsfakhr F et al. Robot Localisation Using UHF-RFID Tags: A Kalman Smoother Approach[J]. Sensors, 2021, 21(3):717.

[41]李俊杰, 黄翔, 李泷杲, 等. 基于人工地标的移动机器人定位与调整技术[J]. 航空制造技术, 2020, 63(05): 80-86.

[42]Wang S,et al. A Lightweight Localization Strategy for LiDAR-Guided Autonomous Robots with Artificial Landmarks[J]. Sensors, 2021, 21.

[43]Lynen S, Achtelik M W, Weiss S, et al. A robust and modular multi-sensor fusion approach applied to MAV navigation[C]//IEEE/RSJ International Conference on Intelligent Robots and Systems. Piscataway, USA: IEEE, 2013: 3923-3929.

[44]高翔. 视觉 SLAM 十四讲: 从理论到实践[M]. 北京: 电子工业出版社, 2017.3.

[45]张保军. 基于旋量理论和李群—李代数的空间机械臂动力学建模与分析[D]. 长沙:

国防科技大学, 2017.

[46]王浩森. 基于深度相机的煤矿井下机器人 SLAM 技术研究[D]. 西安: 西安科技大学,2021.

[47]焦传佳, 江明. 基于 AprilTag 图像识别的移动机器人定位研究[J]. 电子测量与仪器学报, 2021.

[48]RafaeelC.Gonzalez, RichardE.Woods, 数字图像处理（第三版）[M]. 北京: 电子工业出版社, 2016.

[49]RafaeelC.Gonzalez, RichardE.Woods, StevenL.Eddins. 数字图像处理（MATLAB 版）[M]. 北京电子工业出版社. 2014.

[50]阳树洪. 灰度图像阈值分割的自适应和快速算法研究[D]. 重庆: 重庆大学. 2014.

[51]甘志坚, 陈哲, 方俊彬, 等. 复杂光照下 QR 码图像二值化算法研究及应用[J]. 应用光学, 2018, 39(05): 667-673.

[52]WANG J, OLSON E. AprilTag 2: Efficient and robustfiducial detection ［C］．IEEE/ＲSJ International Conference on Intelligent Ｒobots and Systems(IＲOS), 2016: 4193-4198．

[53]袁野. 摄像机标定方法及边缘检测和轮廓跟踪算法研究[D]. 大连: 大连理工大学,

2002.

[54]李莉, 王绪国. 局部二值模式耦合双阈值 LM 优化的火焰图像边缘检测算法[J]. 电子测量与仪器学报, 2019, 33(06): 9-17.

[55]罗文婷, 李中轶, 李林, 等. 基于改进 Canny 算法的道路标线自动识别及定位[J].成都: 西南交通大学学报, 2018, 53(06): 1253-1260.

[56]宋人杰, 刘超, 王保军. 一种自适应的 Canny 边缘检测算法[J]. 南京: 南京邮电大学学报: 自然科学版, 2018, 38(03): 72-76.

[57]樊俊杰, 梁华为, 祝辉, 等. 基于双目视觉的四边形闭环跟踪算法[J]. 机器人, 2015,37(06): 674-682.

[58]孙丽君, 李石, 陈天飞. 基于仿射变换和均值像素法的环状编码点鲁棒识别[J]. 仪器仪表学报, 2021, 42(09): 71-80.

[59]章权兵, 韦穗, 罗斌, 等. 基于射影变换模型的图像特征点集配准[J]. 计算机工程与应用, 2004, 40(34): 42-44.

[60]席志红, 李爽, 曾继琴, 等. 一种改进的 PnP 问题求解算法研究[J]. 应用科技, 2018,45(04): 56-60.

[61]姚伟, 徐骏善, 汪惠芬. 基于视觉导航 AGV 的视觉定位系统设计[J]. 制造业自动化,2020, 42(11): 18-22.

[62]焦传佳, 江明. 基于 AprilTag 图像识别的移动机器人定位研究[J]. 电子测量与仪器学报, 2021, 35(01): 110-119.

[63]Rafael Muñoz-Salinas, et al. UcoSLAM: Simultaneous localization and mapping by fusion of keypoints and squared planar markers[J]. Pattern Recognition, 2020, 101(C).

[64]Jun Wang, Yasutake Takahashi, Antonio Lazaro. Particle Smoother-Based Landmark Mapping for the SLAM Method of an Indoor Mobile Robot with a Non-Gaussian Detection Model[J]. Journal of Sensors, 2019.

[65]杨梦佳. 基于惯导与双目视觉融合的 SLAM 技术研究[D]. 西安: 西安科技大学,

2020.

[66]Volkan Sezer, Ziya uygar Yengin. A NOVEL DATA ASSOCIATION TECHNIQUE TO

IMPROVE RUN-TIME EFFICIENCY OF SLAM ALGORITHMS[J]. Anadolu Üniversitesi Bilim Ve Teknoloji Dergisi A - Uygulamalı Bilimler ve Mühendislik, 2019(2).

[67]高源东. 移动机器人视觉 SLAM 回环检测技术研究[D]. 南京: 东南大学, 2020.

[68]韩波. 面向动态环境的视觉 SLAM 技术研究[D]. 杭州: 浙江大学, 2021.

[69]Dong R, Liu C, Yang G. TF-IDF based loop closure detection algorithm for SLAM[J].Dongnan Daxue Xuebao (Ziran Kexue Ban)/Journal of Southeast University (Natural Science Edition), 2019, 49(2): 251-258.

[70]Abbas Syed Muhammad, Aslam Salman, et al. Analysis and Improvements in AprilTag Based State Estimation[J]. Sensors (Basel,Switzerland), 2019, 19(24).

[71]Shahrizal Saat, WN Abd Rashid, MZM Tumari, et al. HECTORSLAM 2D MAPPING FOR SIMULTANEOUS LOCALIZATION AND MAPPING (SLAM)[J]. Journal of Physics:

Conference Series, 2020, 1529(4).

[72]季秀才. 机器人同步定位与建图中数据关联问题研究[D]. 长沙: 国防科学技术大学,2008.