Date of Award


Degree Name

MS in Aerospace Engineering


Aerospace Engineering


Dr. Eric Mehiel


As Unmanned Aerial Vehicles (UAVs) are integrated into the national airspace to comply with the 2012 Federal Aviation Administration Reauthorization Act, new civilian uses for robotic aircraft will come about in addition to the more obvious military applications. One particular area of interest for UAV development is the autonomous cooperative control of multiple UAVs. In this thesis, a decentralized leader-follower control strategy is designed, implemented, and tested from the follower’s perspective using vision-based localization.

The tasks of localization and control were carried out with separate processing hardware dedicated to each task. First, software was written to estimate the relative state of a lead UAV in real-time from video captured by a camera on-board the following UAV. The software, written using OpenCV computer vision libraries and executed on an embedded single-board computer, uses the Efficient Perspective-n-Point algorithm to compute the 3-D pose from a set of 2-D image points. High-intensity, red, light emitting diodes (LEDs) were affixed to specific locations on the lead aircraft’s airframe to simplify the task if extracting the 2-D image points from video. Next, the following vehicle was controlled by modifying a commercially available, open source, waypoint-guided autopilot to navigate using the relative state vector provided by the vision software. A custom Hardware-In-Loop (HIL) simulation station was set up and used to derive the required localization update rate for various flight patterns and levels of atmospheric turbulence. HIL simulation showed that it should be possible to maintain formation, with a vehicle separation of 50 ± 6 feet and localization estimates updated at 10 Hz, for a range of flight conditions. Finally, the system was implemented into low-cost remote controlled aircraft and flight tested to demonstrate formation convergence to 65.5 ± 15 feet of separation.