Abstract: Routine inspection is an important task in the operation and maintenance of FAST (Five-hundred-meter Aperture Spherical radio Telescope). The telescope consists of a large number of components. Searching for components by their numbers during the inspection is time-consuming and laborious. Solving the position and navigation problem in inspection can help maintenance personnel find fault points quickly. It is important for improving the inspection efficiency and ensuring the telescope runs normally. Given that the satellite signals are blocked by FAST reflector, the inertial navigation is adopted and a set of pedestrian inertial navigation algorithm is designed based on pedestrian foot-mounted IMUs (Inertial Measurement Units). The algorithm follows the Zero Velocity Update (ZUPT) algorithm framework, fuses the data of two IMUs of two feet, and uses the magnetic field information to correct the heading error. The effectiveness of the algorithm was tested in regular trajectories, and the horizontal positioning error was within 3% of the walking distance. Every actuator has accurate coordinates and can be used as marker points to correct the position of inspectors when they pass. In the irregular trajectory of the inspection situation, combined with the control points along the way, the positioning error is less than 11 meters. The precision of algorithm can meet the positioning needs in inspection and support the application of the positioning system for inspection in the future.