Eagle Eye

Abstract

Autonomous drones currently have limited collision resolution capabilities. The top of the line models are able to autonomously avoid large, static objects. However, no such facilities exist in avoiding drone on drone collision. Having every drone individually search for drones around it to avoid collisions is very taxing on the battery and thus not practical. With the imminent rise of autonomous drone usage in package delivery and various other applications, it will be paramount to ensure that drones do not crash into each other. In order to solve this dilemma, we propose to create a decentralized system of controllers attached to the corners of buildings that will act like traffic cops for the autonomous drones. We are focusing on this building corner intersection because part of the drone’s sight is obscured by the building. Thus, it won’t be able to see drones flying on the other side of the building and will have less time to react to a potential collision. Each controller uses a reservation system to reserve voxels that correspond to a drone’s flight path. Drones approach the intersection, reach out to the controller, then provide their desired flight path through the intersection and a couple of their next waypoints. The controller receives the flight path and then tries to reserve the path for the drone. If the requested path does not have any intersection with another drone’s path that is already reserved, the controller reserves that path and then tells the drone to go ahead. If the requested path does have an intersection with another drone’s path, the controller replies no to the drone who requested. The drone then has the option of waiting until the path is free (and keep pinging the controller until it receives a yes response) or deciding to take an alternative path and request the new path from the controller. William