【作者】 王超；

【导师】 穆莉莉；

【作者基本信息】 安徽理工大学 ， 机械工程， 2024， 硕士

【摘要】 受煤层开采、山体压力、地壳岩层运动等因素影响,煤矿井下巷道面临各种扰动荷载动力破坏效应,极易引起巷道变形,甚至塌陷,造成重大安全事故。由于煤矿井下工作环境恶劣,以机器代替人工,面向煤矿井下巷道健康状态的新型智能巡检技术需求尤为迫切。目前巷道形变监测主要采用人工勘察、全站仪、布压力传感器或轨道式机器人。煤矿井下巷道距离长,采用全站仪检测劳动强度大,巡检效率低;布传感器法无法做到全域连续测量,成本也高;轨道式机器人受到轨道布局和自身视角的限制等问题也无法做到对巷道及设备的全面巡检,这与实现煤矿井监控无死角、无盲区的目标相悖。为解决巷道巡检存在的这些问题,本文设计了一种无轨轮式巡检机器人,旨在更加灵活地完成井下巷道的巡检任务,文章主要研究巡检机器人的SLAM建图及路径规划导航等关键技术。具体研究内容如下:(1)研制了巡检机器人平台。选用2D激光雷达RPLIDAR A3进行环境建图与机器人的自身定位、16线激光雷达RS-Helios-16P进行三维建图以补偿二维建图的缺陷;选用高性能的YCT-080-i7-1165G7工控机来提高海量点云数据的处理能力,通过ROS操作系统控制机器人实时完成导航与巡检任务;同时,机器人还搭载有深度相机、里程计、IMU、环境检测等多种辅助传感器,进一步提高建图精度与巡检能力。(2)进行了Gmapping和Karto SLAM二维SLAM建图算法的效果对比研究。选定性能更优异的Karto SLAM作为本文巡检机器人二维建图算法。井下巷道环境复杂、非结构化,为保证给机器人提供更加丰富完整的环境信息,使用了16线激光雷达运行轻量级的三维SLAM建图算法Le GO-LOAM构建了环境三维点云地图,再通过降维三维点云地图来优化二维环境栅格地图,得到了特征更加丰富的环境地图;最后对代价地图和膨胀系数进行了建模,以提高机器人地图的安全系数,为定位导航功能提供保障。(3)建立了全局路径与局部路径规划算法的融合模型。确定了以AMCL为机器人的定位算法,通过对全局A*算法的启发函数、权重系数和路径平滑等方面进行优化处理来提高全局路径规划效果,优化后算法在仿真实验中效率明显提升,路径生成的时间相比优化前缩短一半以上;再通过局部路径规划算法TEB来实现机器人的实时避障功能,仿真环境中机器人在全局以及局部路径规划器的共同作用下能够实现稳定导航。(4)开展了巡检机器人实验平台对多特征复杂场景的建图与路径规划实验研究。结果表明,本文融合算法建图平均误差为1.259%,障碍物尺寸检测误差为2%～4%,机器人在全局融合局部路径规划算法共同作用下以平滑的路径到达目标处,并且在移动中实时避开动态障碍及行人的影响,多次实验中实时避障的成功率保持在96%以上,为实现煤矿井下巡检工作由机器人代替人工提供了技术支撑。图 [78] 表 [13] 参 [93]更多还原

【Abstract】 Affected by coal mining,mountain pressure,crustal rock movement and other factors,coal mine underground roadway faces a variety of disturbing load power damage effect,which is very easy to cause roadway deformation,or even collapse,resulting in major safety accidents.Due to the harsh working environment of underground coal mines,the demand for new intelligent inspection technology for the health status of underground coal mine roadways is especially urgent,replacing manual labor with machines.At present,the roadway deformation monitoring mainly adopts manual survey,total station,cloth pressure sensor or track robot.Coal mine tunnel distance is long,the total station detection labor intensity,inspection efficiency is low;cloth sensor method can not do the whole area of continuous measurement,the cost is high;rail robot by the track layout and its own perspective limitations and other issues can not be done on the roadway and the comprehensive inspection of equipment,which is contrary to the realization of the goal of the coal mine mine well monitoring without dead ends,no blind zones.In order to solve these problems of roadway inspection,this paper designs a trackless wheeled inspection robot,aiming to more flexibly complete the inspection task of underground roadway,the article mainly researches the inspection robot’s SLAM mapping and path planning navigation and other key technologies.Specific research content is as follows:(1)The inspection robot platform is developed.The 2D LIDAR RPLIDAR A3 is selected for environment mapping and robot’s own localization,and the 16-line LIDAR RS-Helios-16 P is used for 3D mapping to compensate for the defects of 2D mapping;the high-performance YCT-080-i7-1165G7 industrial computer is selected to improve the processing capability of massive point cloud data,and the robot is controlled by ROS operating system to complete the navigation and inspection tasks in real time;at the same time,the robot is controlled by ROS operating system to complete the navigation and inspection tasks in real time;at the same time,the robot is controlled by ROS operating system.The robot is controlled by the ROS operating system to complete the navigation and inspection tasks in real time;at the same time,the robot is equipped with a variety of auxiliary sensors,such as depth camera,odometer,IMU,environment detection,etc.,to further improve the accuracy of mapping and inspection capabilities.(2)A comparative study of the effects of Gmapping and Karto SLAM 2D SLAM mapping algorithms is conducted.Karto SLAM,which has better performance,is selected as the 2D mapping algorithm of the inspection robot in this paper.The underground tunnel environment is complex and unstructured,in order to ensure that the robot is provided with richer and more complete environmental information,the16-line Li DAR was used to run the lightweight 3D SLAM mapping algorithm Le GOLOAM to construct a 3D point cloud map of the environment,and then optimize the2 D environment raster map by downscaling the 3D point cloud map to obtain a richer environment map with richer features;finally,a comparison of the cost map and the expansion coefficient were modeled to improve the safety coefficient of the robot map for the localization and navigation function.(3)The fusion model of global path and local path planning algorithm is established.AMCL is identified as the robot’s localization algorithm,and the global path planning effect is improved by optimizing the heuristic function,weight coefficient and path smoothing of the global A* algorithm,and the efficiency of the optimized algorithm is significantly improved in the simulation experiments,and the time of path generation is shortened by more than half compared with that before the optimization;and then the real-time obstacle avoidance of the robot is realized by the local path planning algorithm TEB,and the robot in both global and local paths is optimized by the local path planning algorithm in the simulation environment.In the simulation environment,the robot can realize stable navigation under the joint action of global and local path planners.(4)The experimental platform of inspection robot is carried out for the experimental research of multi-featured complex scene mapping and path planning.The results show that the fusion algorithm in this paper builds a map with an average error of 1.259%,and the error of obstacle size detection is 2%～4%,and the robot arrives at the target with a smooth path under the joint action of global fusion and local path planning algorithm,and avoids the dynamic obstacles and the influence of pedestrians in the movement in real time,and the success rate of real-time obstacle avoidance is maintained at more than 96% in many experiments,which provides technical support for the realization of the inspection of the underground coal mines by the This provides technical support for the realization of underground inspection by robots instead of manual labor.Figure [78] Table [13] Reference [93]更多还原