In some large hydropower stations, the isolated phase enclosed bus (IPB) can only check the equipment status through preventive tests, it is difficult to visually inspect the appearance. To solve the problem of difficult inspection of annular pipelines in such confined and narrow spaces, a wheeled climbing robot that can move in horizontal and vertical segments was proposed. By comparing and analyzing the movement and attachment methods of the climbing robot, and referring to the actual working conditions, the wheeled thrust attachment method was selected. Also, the force analysis of the overall structure of the robot was carried out. The experiments were carried out in the simulated IPB pipeline. The results show that the inspection robot has good adaptability to the IPB pipeline. In addition, a point cloud registration method based on cylindrical coordinates is designed, which solves the problem that the map cannot be correctly constructed in the pipeline, and the calculation time is better than lidar odometry and mapping (LOAM) and iterative closest point (ICP) algorithms, which expands the application range of lidar point cloud mapping.