Unconventional resources that contain shale gas have become significant sources of natural gas. However, unconventional resources are present in complex and horizontal layers. Since conventional drilling methods such as vertical drilling cannot reach complex and horizontal regions, directional drilling, which can change the drilling direction to the desired direction, is required. Underground localization is a major technology for controlling the directional drilling system and path following. In particular, accurate underground localization is essential to increase the production volume and efficiency.
Underground localization has several impediments. For example, the underground environment is a GPS-denied area. Moreover, the underground environment is lightless and narrow and also does not transmit radio waves well. Therefore, the sensors that are used in general localization cannot be practically used in the underground environment. In order to overcome these impediments, inertial measurement unit (IMU) have been widely used in conventional methods for underground localization. There are various methods to mitigate the noise of an IMU or to reduce accumulated errors. However, since there are vibrations and distortion of the geomagnetic field in the drilling environment, the results of these methods are thus easily affected by the external environment. Other papers have proposed localization methods using an artificial magnetic field. These approaches, however, can only measure the relative pose within a short range. Therefore, an innovative underground localization method that can correct the accumulated error and is robust to the external environment is required.
In response, a novel underground localization using the magnetic field distribution is proposed in this study. To minimize the accumulated errors, simultaneous localization and mapping (SLAM) framework is used in this study. In particular, pose graph SLAM method uses a concise graph structure that consists of the nodes represent the poses of the system and the constraints stand for the relative positions between nodes. To detect the constraints, the proposed algorithm utilizes the attributes of the underground magnetic field and directional drilling. First, different soil constituents produce different magnetic field anomalies. Therefore, the region can be distinguished by the magnetic field anomalies. Second, the drilling system repeats forward and backward movements in the borehole. Since this property enables the drilling system to revisit the previous path, the magnetic field anomalies can be re-measured. To find correspondence between the magnetic field anomalies, concurrent normalized cross-correlation and magnetic field vector matching is proposed in this study. Furthermore, to estimate the variation of the orientation between nodes, orientation estimation using least squares is proposed. Using these results, the constraints is detected. Finally, to optimize the path, pose graph optimization is used.
To verify the performance and the applicability of the proposed algorithm, various simulations and field tests are performed. The sensor system for the proposed algorithm is also designed. The results show that the proposed algorithm estimates the 6-degree of freedoms (DOF) pose with high accuracy regardless of the drilling environment.