Abstract: In order to solve problems of difficult positioning and low precision in positioning method of underground mobile robots, a positioning method of underground robot based on strapdown inertial navigation and odometer was proposed. The method uses Kalman filter to perform initial alignment for strapdown inertial navigation to determine initial coordinates of the positioning, and obtain initial attitude transformation matrix. Then, it uses strapdown inertial navigation to calculate position of the robot independently, and uses the speed information output by the odometer and the real-time attitude transformation matrix output by strapdown inertial navigation to calculate the dead reckoning, so as to obtain the position information of the robot again. In order to reduce influence of the accumulated error on the strapdown inertial navigation, it adopts odometer and strapdown inertial navigation to construct dead reckoning system, and uses Sage-Husa adaptive filtering to design combined positioning algorithm to select errors as the system state. After filtering calculation and correction, the accurate position information of the robot is obtained. The experimental results show that the method can realize real-time positioning of the robot and can effectively reduce influence of the accumulated errors of strapdown inertial navigation. And the method has higher positioning accuracy. After the robot moves 4.3 m in Y direction and 0.25 m in Z direction, the positioning error in Y direction is 0.25 m and the positioning error in Z direction is 0.005 m.