Abstract: Aiming at the problem of low pose accuracy and poor measurement stability of roadheader caused by single position measurement method of roadheader in coal mine, a combined positioning method of roadheader based on machine vision and optical fiber inertial navigation is proposed. Taking the three-laser pointer fixed on the roof of the roadway as the characteristic target, a visual measurement system of the body pose of the roadheader based on the three-laser target is constructed. The airborne explosion-proof camera is used to collect the laser target, and the spot center position of the laser pointer is obtained by image processing. Combined with the visual measurement model of the body pose of the roadheader and the spatial coordinate transformation, the pose data of the roadheader are obtained. At the same time, the position and attitude information of the roadheader are obtained in real time by using the fiber optic inertial navigation installed on the roadheader fuselage. The Lagrange three-point interpolation method is used to register the visual pose data and the inertial pose data in time. After processing, the Kalman filter fusion algorithm is used to fuse the visual measurement data and the fiber optic inertial navigation data. The pose data after the fusion of the two is used to compensate and correct the cumulative error of the inertial navigation, and the output validity of the sensor of the measurement system is judged, which eliminates the cumulative error caused by the long-term integration operation of the inertial navigation and the short-term occlusion of the visual target and the influence of data loss on the fusion result, so as to obtain the real-time accurate pose data of the roadheader. The visual measurement platform of roadheader pose is built in the laboratory to evaluate the visual positioning method. Within the measurement distance of 60 m, the maximum position error is within 35 mm, and the maximum attitude angle error is within 0.5°. At the same time, the fusion positioning of vision and inertial navigation of roadheader was tested in the industrial field of underground coal mine. The results show that the fusion positioning method of vision and inertial navigation can overcome the influence of complex working conditions in underground mine and make up for the deficiency of single measurement method. The positioning error of roadheader body is within 30 mm, and the attitude angle error is within 0.5°, which meets the requirements of roadway construction accuracy.