We were inspired by futuristic films in which certain personalities would walk freely while being automatically tracked and live-streamed to the rest of the world by flying machines.
What it does
We managed to establish intuitive control of the quadracopter with the help of the leap-motion. We also used a raspberry pi to broadcast a live video feed.
How we built it
We started with the idea of a project involving a drone and let the idea grow to involve not only control, but also image processing and live-feed broadcasting.
Challenges we ran into
The AR Parrot was running on a firmware which was incidentally highly incompatible with most of the major APIs. We were confident that we would be able to get a real-time video stream from the drone's camera, but requesting the data caused the firmware to become unstable and crash. We therefore had to resort to alternate video sources and that's where we included the raspberry-pi.
Accomplishments that we're proud of
Successfully setting up a web server with the raspberry-pi, establishing comfortable drone control through the leap motion and still-image face recognition and tracking.
What we learned
How to control a parrot drone trough alternate means, how to use the leap motion, how to broadcast live video from a raspberry-pi and setting up an HTTP server in NodeJS, as well as performing various experiments with google carboard and the occulus rift.
What's next for Flying Cameraman
We'd love to analyze and work much more in-depth with the API to establish full communication with the parrot drone and its peripherals. Once this is done, we could use the face recognition as input to a basic tracking AI that would follow a subject around in a 3D environment.