Toward high-quality magnetic data survey using UAV: development of a magnetic-isolated vision-based positioning system

Precise positioning is a prerequisite for airborne high-quality magnetic data surveying. Modern airborne magnetic surveying platforms are usually composed of two parts: an aerial vehicle and a suspended magnetometer assembly. This configuration poses a challenge for georeferencing of data since the sensitivity of the magnetic sensors prohibits the placement of electronic positioning sensors directly on the assembly. This work focuses on the development of a positioning approach using computer vision and sensor fusion, which provides a globally referenced position, as well as preserving the electromagnetic isolation for the magnetic sensors. For the sake of magnetic isolation, the proposed approach estimates the position by fusing the relative position between the UAV and the magnetic sensor as well as the global position of the UAV itself. Compared with other solutions, the proposed solutions present two main advantages: (1) It is passive and provides excellent magnetic isolation to the sensitive magnetic sensor; (2) it can continue to provide the position of the magnetometer even if the sensor is in certain environments where direct positioning using global navigation satellite system or visual odometry would fail. We present the complete solution, including the temporal/spatial alignment of the GNSS/INS system and camera, front-end vision-based detection and tracking, and back-end position estimation and optimization. For performance evaluation, we use a commercial GNSS/INS system on the assembly as a reference and then compare the result of the proposed method with respect to the reference. The results from several real flight tests show that the vertical and horizontal accuracy of the proposed method is at centimeter level and decimeter level, respectively, which comes close to the performance of the reference system.