钢罐道作为矿井提升容器的导向限位装置，其自身表面受损程度、与罐道梁的安装状态以及每段罐道间的位置关系等会直接影响提升容器的安全运行。目前钢罐道的巡检方式仍然为人工巡检，存在安全性差、效率低、漏检率高等问题。为提高立井钢罐道的巡检效率及安全性，提出了一种新型的磁轮驱动巡检机器人模型，并试制了原型样机。首先，根据立井钢罐道实际巡检工况的功能需求分析，对机器人进行结构设计，并研究了磁轮驱动下的可靠运行条件，进而对磁轮的设计尺寸进行优化；其次，在扩展卡尔曼滤波（EKF）方法的基础上，借助加速度计、陀螺仪和磁传感器（合称MARG传感系统）对所设计的机器人进行了姿态估计，并运用反步法对机器人在罐道面上的高效运行进行了步态规划；最后，通过仿真和试验的方法对所设计的磁轮驱动巡检机器人进行了验证。结果表明：① 所提出的姿态估计方法可以实现对机器人当前姿态的准确估计，偏差浮动范围保持在±0.1°内；② 所设计的巡检机器人可以实现直线和圆轨迹的快速、准确跟踪，当跟踪时间达到3 s时，局部位置误差趋于稳定并收敛到0；〖JP2〗③ 所设计的巡检机器人能够进行壁面跨越，当左右两侧舵机转速为0.1 m/s，〖JP〗转弯半径为49.5 mm时，可实现0°～90°的转弯，在转弯初始阶段会出现一定的偏差，当偏差达到1 mm左右时趋于稳定。

As the guiding and limiting device of mine hoisting container in mines, the damage degree of the steel cage guide’s surface, the installation status of the steel cage guide beam and the position relationship between each cage guide will directly affect the safe operation of hoisting container. At present, there are some problems in manual inspection of steel cage guide, such as poor safety, low efficiency and high rate of missing inspection. In order to improve the inspection efficiency and safety of steel cage guide,  a novel inspection robot model driven by magnetic wheel was proposed, and a prototype was trial produced. Firstly, according to the functional requirement analysis of the actual inspection condition of the shaft steel cage guide, the structure of the robot was designed, and the reliable operation conditions under the magnetic wheel drive were analyzed, and then the design size of the magnetic wheel was optimized. Secondly, based on the extended Kalman filter (EKF) method, the attitude of the designed robot was estimated with the help of accelerometer, gyroscope and magnetic sensor (collectively called MARG sensor system), and the gait planning for the efficient operation of the robot on the tank pavement was carried out by using the backstepping method. Finally, the performance of the robot was verified by simulation and experiment. The results show that: ① the proposed attitude estimation method can accurately estimate the real-time attitude of the robot, and the deviation floating range is kept within ± 0.1°; ② the designed inspection robot can realize fast and accurate tracking of straight line and circle tracks, and when the tracking time reaches 3 s, the local position error tends to be stable and converges to 0; ③ the designed patrol robot can realize the wall spanning movement, and when the rotational speed of the left and right steering engines is 0.1 m/s and the turning radius is 49.5mm, it can realize the turning of 0°-90°. A certain deviation will occur at the initial stage of the turning, and it will become stable when the deviation reaches about 1 mm.