Date of Award
Master of Science
Department of Electrical and Computer Engineering
Brett J. Borghetti, PhD
The US Air Force has been increasing the use of automation in its weapon systems to include the remotely piloted aircraft (RPA) platforms. The RPA career field has had issues with poor pilot retention due to job stressors. For example, RPA operators spend a lot of time and attention surveilling a suspect on the ground for many hours, so adding automation to this activity could help improve pilot retention. The research problem in this thesis attempted to automate the process of observing a ground target. This thesis presents a method termed conic ray tracing for determining visibility and occlusion of a ground target from locations in the airspace represented by 3D point cloud data. This conic ray tracing method uses 3D points representing a scene to trace rays and then using a matrix formulation of the dot product to compute the angles between every ray to the points representing airspace and every ray to the points representing the ground scene. Whether the angle is inside or outside a fixed-angle cone determines occlusion or visibility respectively. The method was tested on 3D point clouds generated from Structure from Motion using real-world imagery data collected from an MQ-9 Reaper ight test. Because the truth data was not available, the results from conic ray tracing were compared to a reference created by using true graphical ray tracing on a surface reconstruction of the original point cloud scenes. When compared to the reference, the conic ray tracing method averaged 0.81 accuracy over the eight cases studied, and the computational runtime was on average 19x faster than the algorithm for computing the reference results from true ray tracing. Limitations and future work were discussed.
DTIC Accession Number
Cho, Henry, "Comparison of Conic Ray Tracing for Occlusion Determination on 3D Point Cloud Data" (2021). Theses and Dissertations. 4888.