3D spatial mapping of roadways based on multi-sensor fusion

Based on the insufficient intelligent perception capability of coal mine heading face in China, this paper proposes a three-dimensional spatial mapping method of coal mine roadways based on multi-sensor fusion. This method utilizes 3D sensors such as LiDAR to capture and preview the environmental information of the coal mine roadways, and constructs a real-time autonomous mapping model to provide spatial perception information for large-scale coal mine heading equipment. By optimizing the Extended Kalman Filter algorithm based on the attitude perception theory and implementing the multi-sensor fusion technology, the method achieves fast updates of global maps and high-precision mapping in the geometrically degraded mine heading environment. The main contribution of this study lies in the fusion of multiple sensor data to construct a high-precision underground three-dimensional map and improve the intelligent perception capability and autonomous operation level in the mine heading environment. Firstly, some advanced LiDAR sensors are used to capture point cloud data within the mine heading face, extract point cloud features, and establish inter-frame feature mapping models to create a centimeter-level underground three-dimensional map. Meanwhile, by fusing data from LiDAR with inertial navigation using an iterated error state Kalman filter, the robustness of the system is improved, ensuring the accuracy and stability of the underground three-dimensional map. Additionally, this paper considers the influence of factors such as underground dust and water mist on the accuracy of three-dimensional spatial mapping of mine roadways. The Kalman filter gain is optimized to adjust the weight of LiDAR odometry, enhancing the environmental adaptability of three-dimensional spatial mapping under low dust concentration conditions. Experimental results show that comparing to the LeGO-LOAM and LINS, the proposed three-dimensional spatial mapping method outperforms in terms of trajectory estimation accuracy, point cloud quantity, and point cloud density in mapping the environment. The method presented in this paper can be utilized to achieve high-precision and fast-updating three-dimensional maps for autonomous mine heading equipment, thus improving the intelligence level of coal mine heading face and providing a data support for the intelligent perception and autonomous operation in the mine heading environment.