iFollow uses Remote Image Processing to search for targets and remotely, autonomously control a quadcopter. The video feed is processed and the precise movements are calculated and sent to the quadcopter. We are a team of two: George Matus (High School Junior, this is his first hackathon), and Harris Christiansen (Freshman Purdue CS).
The idea behind "iFollow" is one that many fantasize about when they think "Quadcopter". Imagine a Quadcopter that can follow you around. Maybe your are mountain biking, skiing, or rock climbing even, and the quadcopter autonomously follows you around, filming you. GoPro to the next level!
There are many other uses for the technology, including easy target tracking (click a target to follow), and search+rescue (autonomous flight searching for a target).
To make this project work, especially with the tools we had on hand, was difficult. We brought with us two basic quadcopters, a cheap, low-quality analog video camera, and a video receiver. Using these tools we were able to stream low quality footage from a quadcopter down to a computer to perform image processing. With only one camera, it was very very difficult to judge distance, making the forward/backward movement required for follow-me very difficult. We overcame this obstacle by limiting our target to a human, a subject of a fairly standard, known size. It was also difficult to refine the movement calculations to work satisfactorily to keep the quadcopter successfully following the target at a safe distance.