Combined Positioning Method for Roadheader Based on Multi-sensor Fusion

In view of the complex working conditions of coal mine underground heading machines, which are prone to slipping, lateral shifting, and large-angle deflection during operation, traditional single-sensor positioning methods struggle to simultaneously meet the requirements of high reliability and anti-interference capability. A combined positioning method for heading machines, integrating LiDAR, fiber-optic inertial navigation, and UWB technology, is proposed. Firstly, the tunnel coordinate system, heading machine coordinate system, LiDAR coordinate system, and navigation coordinate system are established, and the attitude transformation and position transformation matrices between multiple coordinate systems are derived. Single-line LiDAR is used to collect point cloud data of the tunnel roof guide rail targets. After filtering and preprocessing, the DBSCAN density clustering algorithm is employed to separate the guide rail, roof, and noise point clouds. The lateral offset and height of the heading machine relative to the tunnel centerline are calculated by fitting the guide rail centerline. By integrating the ranging information from front and rear UWB base stations, the heading angle measured by inertial navigation, and the lateral offset obtained from LiDAR, a body-tunnel geometric relationship model is constructed to achieve real-time calculation of the heading machine's footage. Based on this, static positioning tests and underground heading machine autonomous cutting tests were conducted using the EBZ220S heading machine as the object. The results show that under static conditions, the average error in lateral deviation measurement is 1.26 cm; the average error in height measurement is 1.72 cm; and the average error in footage measurement is 1.73 cm. The combined positioning method can effectively eliminate the footage measurement error caused by the rotation of the heading machine body, with a maximum suppression of 9.83 cm. In the autonomous cutting test of the heading machine, the machine can stably complete autonomous walking, autonomous correction, and full-section cutting. The average error in the size of the formed section is 5.7 cm, meeting the accuracy requirements for intelligent heading and forming in coal mines. The multi-sensor fusion positioning method proposed in this paper significantly improves the positioning accuracy, stability, and engineering adaptability of the heading machine under actual underground cutting conditions, providing reliable technical support for autonomous walking, full-section autonomous cutting, and remote unmanned operations of the heading machine.