Self-location of Unmanned Aerial Vehicle Using Conformal Geometric Algebra

Abstract

Self-location is a major function of unmanned aerial vehicle (UAV). In the absence of satellite navigation signals, a UAV depending merely on a low-accuracy inertial measurement unit (IMU) certainly fails to meet the needs of self-location. This paper introduces a monocular vision-based self-location method of UAV just fitted with a single downward-looking camera. A self-location mathematic model is built in accordance with the collinear relationship among the ground target point, the image point and the optical center of the camera. By taking conformal geometric algebra (CGA) as a mathematic tool, direct operation for geometric elements and geometric relationship is performed along with efficient extension, unary elimination and simplification. Combined with the least square estimation, the spatial location of UAVs can be calculated. Finally, the performance of this approach is verified via a flight test, experimental results show that the UAV location can be obtained with this method, which also embraces significant advantages in the convergence rate and computing time. Furthermore, it can be deployed to embedded devices with limited computing resources, thus it meets the positive real-time requirements of self-location of UAVs.