Optimal Search Strategies for Unmanned Aerial Vehicles

Open Access
DeAngelo, Mark Patrick
Graduate Program:
Aerospace Engineering
Master of Science
Document Type:
Master Thesis
Date of Defense:
Committee Members:
  • Joseph Francis Horn, Thesis Advisor
  • search
  • UAV
  • UAVs
  • UAS
  • optimal
  • unmanned
  • aerial
  • vehicles
  • camera
  • cameras
  • airplane
  • optimization
  • surveillance
This work presents efficient search methods for an unmanned aerial system (UAS) intended for autonomous flight. The objective is to maximize searched area subjected to the limitations of an optical sensor when searching for a ground target. The work is mostly exhibited with simulations, but it also includes actual flight test demonstrations. This paper is arranged by presenting the simplest configurations first, followed by increasing the levels of configuration complexity. First, a fixed, downward-pointing camera mounted to the underside of an airplane is simulated flying straight and level at a constant altitude. Parameters such as airspeed and parallel flight path separation are optimized while restricted by camera intrinsics and the relative size of a ground target. Next, the airplane is allowed to fly with gentle banks to the left and right, throwing the camera’s optical axis to the left and right sides of the flight path. In that case, the optimizer solves for the best airspeed, bank angles, and track spacing. The scenario is then extended to allow the camera to tilt and zoom. Next, a pan-tilt-zoom camera mounted to an airplane is simulated while being subjected to the technical limitations of the camera, the aircraft, and the relative size of the target on the ground. An optimizer solves for the best camera angles, zoom setting, and the flight path separation between parallel sweeps. The MATLAB simulation results presented are specific to the parameters of the hardware utilized in the laboratory; however, the simulation code is written to accept generalized airplane and camera systems. Furthermore, alternative camera gimbals are also simulated. Finally, flight tests involving an airplane with a downward-pointing fixed camera were conducted to compare against the simulation results. The simulation results indicate that a three-axis nonstandard gimbaled camera configuration searches the most ground area within the objective search region. Since performance is measured by computing area searched per second, the gimbaled camera configurations have a speed advantage. Because the airplane is able to fly straight and level at higher airspeeds, the search requires less time to complete. Furthermore, the gimbaled camera allows the airplane to enter a 180° turn sooner because the gimbaled camera can point to the remaining search area during the turn, unlike the fixed camera. However, the airplane performing banking maneuvers with the fixed camera performed better than expected by searching the most total area. Another advantage of the fixed camera configuration includes less complexity and the absence of a protruding gimbal mounted to the fuselage. The fixed camera configuration has challenges though. One of the most critical challenges is precisely following the waypoints that cover the search area. Simulations show ideal results, but the experiments show that achieving those results depend highly on the airplane’s ability to maneuver to those waypoints. Finally, the bank maneuvers required more time for the airplane to sweep the search area, which decreases the performance when measured by area searched per second.