Abstract: According to the boundary information of the coal wall and roof in the 3D laser scanning model of fully mechanized working face, the shearer can automatically adjust the cutting height of the drum to realize the coal precise mining. The existing technology realizes the automatic extraction of the roof line of coal cutting based on the laser point cloud of the working face. But the extraction results cannot be directly applied to the digital automatic coal cutting. In order to solve this problem, the overall scheme of 3D laser scanning modeling for fully mechanized working face is proposed. The key technologies such as boundary extraction of coal wall and roof, target ball detection, point cloud registration and coordinate transformation are studied. The near real-time acquisition of boundary information of coal wall and roof under 3D geological coordinate system is realized. The information can be directly sent to the shearer drum to provide data reference for the next cutting of the shearer. The scanning of the working surface is realized through an inspection robot to obtain the inspection point cloud. Based on the curvature characteristics of the boundary of the coal wall and roof, the string and normal vector method is used to extract the boundary of the coal wall and roof roughly. The angle information between the normal vector of data points and the normal vector of adjacent points is introduced. The obvious intersection points of non coal wall and roof are eliminated through the threshold. As the inspection point cloud and the extracted boundary information are both located in a local coordinate system, the head and tail point clouds and the inspection point cloud are registered through the detection and registration of the positioning target ball. The working face combined point cloud is obtained. According to the 3D geological coordinate and the local coordinate of the positioning target ball, the transformation relation between the coordinates is obtained. The combined point cloud is transformed into the 3D geological coordinate system through coordinate transformation. Therefore, the boundary information of the coal wall and the roof under the 3D geological coordinate system is obtained. The underground industrial test results show that the error of the boundary between the coal wall and roof extracted by 3D laser scanning technology in fully mechanized working face is less than 10 cm. The sampling points with error less than 4 cm account for 50%. The sampling points with error less than 8 cm account for 96.67%.