CTRL

Danny Broyles passes PhD Defense!

Congratulations Danny for successfully defending his PhD thesis.

Congratulations to Danny Broyles for passing his PhD defense! He has developed an altitude-constrained and occlusion-aware path-planning algorithm with guaranteed coverage. His algorithm will enable UAVs to thoroughly search an area of interest, such as finding lost humans in wilderness environments while satisfying practical altitude constraints such as those set by the FAA.

Abstract: Autonomous Unoccupied Aerial Vehicles (UAVs) have transformed how search and rescue (SAR) missions operate, from expensive crewed helicopter flights to low-cost and rapid real-time video feedback over large and difficult-to-access areas. However, in the case of wilderness SAR missions, which are characterized by complex terrains and dense vegetation, ensuring full search coverage is challenging, and current autonomous UAV path planning solutions do not account for occlusion. When supporting SAR operations, UAV operators often rely on manual control to mitigate the effects of occlusion, performing on-the-fly position adjuments to achieve better, less-occluded viewing angles. However, this human-in-the-loop process can be slow, error-prone, and draws operator attention away from other critical tasks, such as image analysis and mission coordination. To address the challenge of occlusion-aware coverage path planning, this dissertation presents VWSGA+, a viewpoint generation and waypoint planning algorithm that guarantees complete coverage in environments with complex terrains with dense vegetation. The VWSGA+ approach encodes occlusion-related characteristics of an environment as a graph problem and generates covering viewpoint sets by solving this graph. Viewpoints are then sequenced to minimize the total distance traveled, and the ordered waypoint set is converted into an executable UAV mission. VWSGA+ builds upon existing viewpoint generation methods by accounting for real-world practical constraints, such as confining the path to a user-defined 3D geofence and the limited flight times of UAVs. The practicality and theoretical guarantees of VWSGA+ are demonstrated in a full-scale simulation and real-world experiments. VWSGA+ is compared against state-of-the-art methods and is shown to be superior in providing complete coverage using battery-efficient paths that respect the 3D boundary. This dissertation takes a step towards advancing autonomous UAV SAR missions that are robust to any wilderness environment and yield high-confidence aerial search coverage, enhancing the search team’s ability to provide this life-saving service to society.