Precise Localization Method for Coal Mine Roadheader Based on the Fusion of Depth Camera and Inertial Navigation Systems

Abstract Addressing the positioning challenges faced by coal mine roadheaders in complex underground environments, this paper proposes a precise positioning method based on the fusion of depth cameras and Inertial Navigation Systems (INS). To overcome the difficulties of visual attitude detection and the accumulation of INS errors, the method utilizes depth cameras to capture point clouds of roof bolts in the tunnel. Through point cloud processing and spatial coordinate transformation, the pose data of the roadheader are measured and used to assist the INS in obtaining real-time pose information. By employing an Error-State Kalman Filter（ESKF） to integrate information from both sources, the method provides real-time and precise pose data for the roadheader, achieving accurate positioning in coal mine tunnels. Experiments conducted on a simulated tunneling face demonstrate that within the effective measurement range, the method achieves an average position error of 24.60 mm and a maximum error of 61.04 mm in the lateral (widthwise) direction, and an average position error of 16.06 mm and a maximum error of 43.77 mm in the longitudinal (digging) direction. These results validate that the proposed method effectively meets the high precision and high stability requirements for positioning in complex coal mine tunneling environments.