在机器人技术大力发展的时代背景下，煤矿巡检机器人将朝着小型化、多元化、集成化、智能化的方向发展。路径规划和同时定位与建图（Simultaneous Localization and Mapping，SLAM）是巡检机器人在陌生环境地图构建与自主导航的关键技术。目前煤矿水泵房巡检机器人使用单一传感器时，存在定位精度低、地图信息不完整以及路径规划效率低等问题。因此，对水泵房巡检机器人定位、构建地图和路径规划过程进行改进和优化具有重要的研究意义。

针对煤矿井下环境对巡检机器人高精度定位的需求，提出基于误差状态卡尔曼滤波（Error State Kalman Filter，ESKF），将惯性测量单元与轮式编码器的数据作为状态预测值，采用视觉里程计进行位姿校正，提高机器人的定位精度。针对激光雷达构建地图信息不完整、可靠性差等问题，将深度相机的深度信息与激光雷达的数据融合，并通过激光与视觉双回环检测对地图进行优化，减小累积误差，提高全局地图的完整性。针对传统蚁群算法在路径规划前期搜索盲目性大、搜索时间长、拐点多等问题，提出增强目标启发信息蚁群算法。将目标位置对下一个待选节点的影响进行放大，引入带有权重的距离启发因子，使蚂蚁大概率向目标栅格搜索，提高了路径规划的效率。采用正弦自适应调整信息素挥发因子，增强算法的全局搜索能力，并删除冗余路径，进行路径安全性检查，重新调整路径，从而减少转弯的次数。最终融合增强目标启发信息蚁群算法和动态窗口算法，实现巡检机器人自主导航和动态避障功能。

实验结果表明，本文提出的多传感器融合SLAM算法可以构建信息完整的水泵房二维地图，融合定位方法可以实现高精度定位，定位精度为8.9 cm。提出的增强目标启发信息蚁群算法在30×30 m²和100×100 m²的环境中比多因素蚁群算法在搜算时间上缩短了17%和11%，且规划路径更短更平滑。因此，本文提出的算法可以有效应用于水泵房巡检机器人自主巡检任务中。

In the background of the era of vigorous development of robotics, coal mine inspection robots will develop in the direction of miniaturization, diversification, integration and intelligence. Path planning , simultaneous localization and map are the key technologies for map construction and autonomous navigation of inspection robots in unfamiliar environments. The current coal mine pump room inspection robot has problems such as low positioning accuracy, incomplete map information and low path planning efficiency when using a single sensor. therefore, Therefore, it is important to improve and optimize the process of positioning, building maps and path planning for pump room inspection robots.

In response to the demand for high-precision positioning of inspection robots in underground coal mine environment, error-state Kalman filtering is proposed to improve the positioning accuracy of robots. The data from the inertial measurement unit and the wheel encoder are used as state prediction values, and the visual odometry is used for position correction. To address the problems of incomplete information and poor reliability of the map constructed by LiDAR, the depth information of the depth camera is fused with the data of LiDAR, and the map is optimized by double loopback detection of laser and vision to reduce the cumulative error and improve the integrity of the global map. The enhanced goal-inspired information ant colony algorithm with enhanced target direction is proposed for the problems of traditional ant colony algorithm such as large search blindness, long search time and many inflection points in the pre-path planning stage. The influence of the target position on the next node to be selected is amplified, and the distance heuristic factor with weights is introduced to make the ants search toward the target raster with high probability, which improves the efficiency of path planning. A sinusoidal adaptive adjustment of the pheromone volatility factor is used to enhance the global search capability of the algorithm, and redundant paths are removed, path safety checks are performed, and paths are readjusted so as to reduce the number of turns. Finally, the enhanced goal-inspired information ant colony algorithm and dynamic window algorithm are fused to realize the autonomous navigation and dynamic obstacle avoidance function of the inspection robot.

The experimental results show that the multi-sensor fusion SLAM algorithm proposed in this paper can build a two-dimensional map of the pump room with complete information, and the fusion localization method can realize high-precision localization with a localization accuracy of 8.9 cm. The proposed enhanced goal-inspired information ant colony algorithm reduces the search time by 17% and 11% in 30×30 m² and 100×100 m² environments than the multi-factor ant colony algorithm, and the planning path is shorter and smoother. Therefore, the algorithm proposed in this paper can be effectively used in the autonomous inspection task of pump room inspection robot.

[1] 江新奇, 刘敬玉, 李忠飞等. 煤矿巡检机器人智能传感与控制系统设计研究[J]. 煤炭工程, 2022, 54(01): 171-175.

[2] 葛世荣, 胡而已, 裴文良. 煤矿机器人体系及关键技术[J]. 煤炭学报, 2020, 45(1): 455-463.

[3] 袁静. 煤矿巡检机器人自主建图与导航技术研究[D]. 沈阳: 辽宁工程技术大学, 2021.

[4] 聂珍, 马宏伟. 煤矿巡检机器人巷道气体环境智能检测系统设计[J]. 工矿自动化, 2020, 46(06): 17-22.

[5] 宣鹏程, 孙稚媛, 周东旭等. 煤矿轨道式带式输送机巡检机器人系统设计[J]. 煤矿机械, 2020, 41(05): 1-3.

[6] 杨林, 马宏伟, 王岩等. 煤矿巡检机器人同步定位与地图构建方法研究[J]. 工矿自动化, 2019, 45(09): 18-24.

[7] 齐政光. 基于激光雷达与深度相机数据融合的机器人导航系统研究[D]. 济南: 济南大学, 2022.

[8] 鲍周洋. 煤矿井下巷道全方位巡检轮式机器人设计与研究[D]. 徐州: 中国矿业大学, 2022.

[9] 金书奎, 寇子明, 吴娟. 煤矿水泵房巡检机器人路径规划与跟踪算法的研究[J]. 煤炭科学技术, 2022, 50(05): 253-262.

[10] 许会星. 煤矿自动巡检机器人控制系统研究[J]. 煤炭技术, 2022, 41(08): 241-243.

[11] Kasprzyczak L, Trenczek S, Cader M, et al. Pneumatic robot for monitoring hazardous environments of coal mines[J]. Solid State Phenomena, 2013, 7(41): 120-122.

[12] Gomathi V, Sowmya A S, Avudaiammal P S, et al. Design of an adaptive coal mine rescue[C]. Foundation of Computer Science, 2015, 6(24): 219-229.

[13] Jatsun S, Yatsun A, Savin S. Pipe Inspection parallel-link robot with flexible structure[M]// Adaptive Mobile Robotics. 2017,8(46):713-719

[14] 刘琦. 单轨悬挂式胶带输送机巡检机器人研究与设计[J]. 煤炭学报, 2021, 27(7): 62-63.

[15] 杨春雨, 张鑫. 煤矿机器人环境感知与路径规划关键技术[J]. 煤炭学报, 2022, 30(1): 1-29.

[16] 张力, 马宏伟, 梁艳等. 煤矿四旋翼巡检机器人系统设计[J]. 煤炭工程, 2022, 53(02): 180-185.

[17] 姜俊英, 周展, 曹现刚等. 煤矿巷道悬线巡检机器人结构设计及仿真[J]. 工矿自动化, 2018, 44(05): 76-81.

[18] 周治国, 曹江微, 邸顺帆. 3D激光雷达SLAM算法综述[J]. 仪器仪表学报, 2021, 42(09): 13-27.

[19] 王霞, 左一凡. 视觉SLAM研究进展 [J].智能系统学报, 2020, 15(05): 825-834.

[20] 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.

[21] 王立玲, 朱旭阳, 马东等. 基于点线特征视觉惯性融合的机器人SLAM算法[J]. 中国惯性技术学报, 2022, 30(06): 730-737.

[22] Engel J, Koltun V, Cremers D. Direct sparse odometry[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 40(03): 611-625.

[23] 葛红飞, 李轶然. 基于点边缘特征自适应融合的视觉SLAM算法[J]. 计算机与现代化, 2023(01): 107-112+120.

[24] 余联想, 郑明魁, 欧文君等. 多传感器融合的移动机器人室外激光SLAM算法优化与系统实现[J/OL]. 电子测量与仪器学报,2023, 37(02): 48-55.

[25] 徐晓苏, 李诺, 姚逸卿. 基于快速回环检测的室外环境下激光雷达SLAM算法[J]. 中国惯性技术学报, 2022, 30(06): 716-722.

[26] 杨林, 马宏伟, 王岩. 基于激光惯性融合的煤矿井下移动机器人SLAM算法[J]. 煤炭学报, 2022, 47(09): 3523-3534.

[27] RMur-Artal R, Montiel J. M, Tardós J. D. A versatile and accurate monocular SLAM system[J]. IEEE Transactions on Robotics, 2015, 31(05): 1147-1163.

[28] 蔡睿, 章国宝, 朱宏伟. 基于局部块特征优化的室内高精度激光SLAM方法[J]. 测控技术, 2023, 34(03): 1-8.

[29] Tixiao Shan, Brendan J. Englot, Carlo Ratti. Tightly-coupled lidar-visual-inertial odometry via smoothing and mapping[J]. ICRA. 2021, 5692-5698.

[30] Chou C.C, Chou C. F., Efficient and accurate tightly-coupled visual-lidar SLAM[J]. IEEE Transactions on Intelligent Transportation Systems. 2022, 77-86.

[31] Liu Q, Li R, Hu H, Gu D. Using unsupervised deep learning technique for monocular visual odometry[J]. IEEE Access. 2019, 76–88.

[32] 卢俊颖, 刘键均, 夏益民等. 动态环境下基于深度学习的实时视觉SLAM[J]. 计算机应用, 2022, 42(2): 86-91.

[33] 张凤, 王伟良, 袁帅等. 动态环境下基于卷积神经网络的视觉SLAM方法[J]. 沈阳工业大学学报, 2022, 44(06): 688-693.

[34] An Y, Shi J, Gu D. et al. Visual-LiDAR SLAM based on unsupervised multi-channel deep neural networks[J]. Cogn Computer, 2022, 1496–1508.

[35] J Li, Z Li, Y Feng. et al. Development of a human–robot hybrid intelligent system based on brain teleoperation and deep learning SLAM[J]. IEEE Transactions on Automation Science and Engineering. 2022, 1-11.

[36] 林韩熙, 向丹, 欧阳剑等. 移动机器人路径规划算法的研究综述[J]. 计算机工程与应用, 2021, 57(18): 38-48.

[37] 张松灿. 基于蚁群算法的移动机器人路径规划研究[D]. 洛阳: 河南科技大学, 2021.

[38] 封建湖, 张婷宇, 封硕等. 面向机器人路径规划的改进粒子群算法[J]. 机械设计与制造, 2021(09): 291-294+298.

[39] 陈秋莲, 郑以君, 蒋环宇等. 基于神经网络改进粒子群算法的动态路径规划[J]. 华中科技大学学报（自然科学版）, 2021, 49(02): 51-55.

[40] Wu X., Bai W., Xie Y. et al. A hybrid algorithm of particle swarm optimization, metropolis criterion and RTS smoother for path planning of UAVs[J]. Applied Soft Computing. 2018,73(4):735-747.

[41] Ma, Z Y, Chen J. Adaptive path planning method for UAVs in complex environments[J]. Journals & Books. 2022, 103-114.

[42] 汪四新, 谭功全, 蒋沁等. 基于改进A*算法的移动机器人路径规划[J]. 计算机仿真, 2021, 38(09): 386-389+404.

[43] 田海波, 李陆军, 畅科剑等. 用于无人车路径规划的改进A*算法[J]. 现代制造工程, 2021(11): 63-68+92.

[44] 黄梦涛, 李智伟. 改进ARA*算法的移动机器人路径规划[J]. 计算机工程与应用, 2022, 58(24): 291-297.

[45] Ebendt R, Drechsler R. Weighted A* search-unifying view and application[J]. Artificial Intelligence. 2009, 173(14): 1310-1342.

[46] 许万, 杨晔, 余磊涛等. 一种基于改进RRT*的全局路径规划算法[J]. 控制与决策, 2022, 37(04): 829-838.

[47] 于强, 彭昭鸿, 黎旦等. 基于MI-RRT*算法的路径规划研究[J/OL]. 现代防御技术, 2023, 22(03): 1-9.

[48] 王海芳, 张瑶, 朱亚锟等. 基于改进双向RRT*的移动机器人路径规划算法[J]. 东北大学学报（自然科学版）, 2021, 42(08): 1065-1070+1142.

[49] 杨立炜, 付丽霞, 郭宁等. 多因素改进蚁群算法的路径规划[J]. 计算机集成制造系统, 2022, 34(05), 1-18.

[50] 昝新宇, 张铁峰, 苑津莎. 基于改进蚁群算法的移动机器人火灾救援路径规划方法［J］. 科学技术与工程, 2021, 21(17): 7243-7248.

[51] CHEN G Z, YAN C L, WU Y Y. Path planning optimization of indoor mobile robot based on adaptive ant colony algorithm[J]. Computers and Industrial Engineering. 2021, 56(06): 723-733.

[52] Akka K, Khaber F. Mobile robot path planning using an improved ant colony optimization[J]. International Journal of Advanced Robotic Systems, 2018, 52(05): 1–7.

[53] 李孟锡, 何博侠, 周 俣. 基于A*和蚁群算法的移动机器人多目标路径规划方法[J]. 机械与电子, 2021, 39(06): 61-65+69.

[54] Jianjun Yu, Yang Wang, Xiaogang Ruan. etc al. AGV multi-objective path planning method based on improved cuckoo algorithm[C]// IEEE 4th Advanced Information Technology, Electronic and Automation Control Conference. 2019, 556-561.

[55] Xiao Tang, Tong Zhou, Jie Yu, etc al. An improved fusion algorithm of path planning for automated guided vehicle in storage system[C]// IEEE 4th International Conference on Computer and Communications. 2018, 510-514.

[56] 吴欣. ROS下移动机器人激光雷达地图构建与路径规划研究[D]. 西安: 西安理工大学, 2021.

[57] 李智伟. 室内移动机器人视觉定位与路径规划研究[D]. 西安: 西安科技大学, 2022.

[58] Shen, Y., Dai, Y. Fast automatic differentiation for large scale bundle adjustment[J]. IEEE Access, 2018, 11146–11153.

[59] Youn, W., Andrew Gadsden, S. Combined quaternion-based error state kalman filtering and smooth variable structure filtering for robust attitude estimation[J]. IEEE Access, 2019, 7(29):148989–149004.

[60] Shaukat, N., Ali, A., Javed Iqbal. et al. Multi-sensor fusion for underwater vehicle localization by augmentation of RBF neural network and error-state kalman filter[J]. Sensors, 2021, 21(4):1149-1175.

[61] Piao, J.C., Kim, S.D. Adaptive monocular visual–inertial SLAM for real-time augmented reality applications in mobile devices[J]. Sensors, 2017, 17(11): 2567-2594.

[62] Jeon, J, Jung, S, Lee, E. et al. Run your visual-inertial odometry on NVIDIA Jetson: Benchmark tests on a micro aerial vehicle[J]. IEEE Robotics and Automation Letters, 2021, 30(7): 5332–5339.

[63] M., Yang, S., Van, M., Li. et al. Ant colony optimization algorithms for dynamic optimization: A case study of the dynamic travelling salesperson problem[J]. IEEE Computational Intelligence Magazine, 2020, 23(1): 52–63.

[64] 陈勇, 范平清, 袁涛. 多因素改进势场蚁群算法的路径规划[J]. 云南大学学报(自然科学版), 2022, 44(04): 718-728.

[65] 王豪杰, 马向华, 代婉玉等. 改进DWA算法的移动机器人避障研究[J]. 计算机工程与应用, 2022, 59(06): 326-332.

[66] 魏立新, 张钰锟, 孙浩等. 基于改进蚁群和DWA算法的机器人动态路径规划[J]. 控制与决策, 2022, 37(09): 2211-2216.