In recent years, remote controlled helicopter flight has been one of the fastest growing hobbies worldwide. A leading factor in the rising interest is the ability to automate flight by using an onboard computer and GPS receiver to navigate through virtual waypoints without human interaction. At altitudes above trees, buildings, and power lines, the system gathers information from dozens of GPS satellites, making its location information remarkably accurate; however, below these obstructions or indoors, the GPS location can be dangerously inaccurate. My project presents a solution to this problem by using vision-based tracking, rather than GPS, to land a remote controlled helicopter on a landing pad. Additionally, by using inexpensive, commonly available hardware, as well as open-source computer vision software, this augmentation can be easily passed along to other hobbyists without breaking the bank.
Eye in the Sky: Augmenting Automated Flight with Computer Vision