EPFL | Media & Design Lab | Master Semester project
Immanuel Koh (Supervisor)
Jeffrey Huang (Professor)
Personal Interactive Autonomous Drone: Proposal of a Supervised Implementation
Today, using drones as a work tool is common business. Each generation of drones becomes cheaper, more independent, less supervised, allowing nearly anyone to use them. In this project we implement a supervised autopilot for a Parrot Bebop 2 drone in an Android application1 . With Parrot SDK and the drone’s real time single perspective view, computer vision algorithms control the drone’s behavior. From color blob detection and tracking, we teach the drone to recognize predefined subjects and to react accordingly using predefined procedures. With a small interaction experiment, we discuss some limitations of our implementation and some possible future improvements. With this first step of interaction, we want to seek possibilities for drones to replace personal assistants or to become an autonomous support for everyday life.