KATHERINE JAMES

Developing an unmanned aerial vehicle capable of

autonomous collision avoidance

Supervisors: Prof. Karen Bradshaw and Mr Anthony Sullivan

Abstract

The use of unmanned aerial vehicles or drones is becoming widespread in society.

The Parrot AR Drone 2.0 is a commercially available drone, created as a toy but an ideal candidate

for experimentation. The standard AR Drone has no obstacle avoidance built into its groundstation

controlled API, Node-ar-drone, for set flight routines. This thesis aims to extend the AR Drone's

abilities to include NodeJS based object avoidance by making use of additional ultrasonic sensors and

associated hardware to allow for distance detection. Three ultrasonic sensors are interfaced to

the drone as well as a microcontroller and the resulting distance data is sent across to the groundstation

via an ESP8266 WiFi module. If an object is detected closer than a certain threshold, an object

avoidance routine is entered into, upon conclusion of which the drone's on-board magnetometer is

used to find the drone's initial heading, allowing the drone to continue on its interrupted

forward flight.