The United States Air Force relies upon aerial refueling to fulfill its missions. Unmanned aerial systems (UAS) and remotely piloted aircraft (RPA) do not currently have access to this capability due to the lack of an on-board pilot to safely maintain a refueling position. This research examines stereo vision for precision relative navigation in order to accomplish the Automated Aerial Refueling (AAR) task. Previous work toward an AAR solution has involved the use of Differential Global Positioning (DGPS), Light Detection and Ranging (LiDAR), and monocular vision. This research aims to leverage organic systems in future aircraft to compliment these solutions. The algorithm presented here generates a point cloud from the disparity between stereo camera images. The algorithm then ts the point cloud to a digital model using a variant of iterative closest points (ICP). The algorithm was tested using simulated imagery of an F-15E rendered in a 3D modeling environment. Experimental results showed a significant increase in accuracy as the receiver aircraft approached the tanker aircraft, reporting accuracies within +/-10cm at distances less than 17m. The algorithm's ability to transition to the real world was validated qualitatively using a 1:7 camera and model setup.
Precision Relative Positioning for Automated Aerial Refueling from a Stereo Imaging System
2015
77 pages
Report
Keine Angabe
Englisch
Combining Stereo Vision and Inertial Navigation for Automated Aerial Refueling
AIAA | 2017
|Combining Stereo Vision and Inertial Navigation for Automated Aerial Refueling
Online Contents | 2017
|