Examination Committee

Prof Shaojie SHEN, ECE/HKUST (Chairperson)
Prof Zexiang LI, ECE/HKUST (Thesis Supervisor)
Prof Lu FANG, ECE/HKUST
 

Abstract

Quadcopter research and application have drawn great attention in the past years. Recent successful commercialization of quadcopter make it become a popular tool applied in areial photography, surveillance, Precision Farming. There is no doubt that in these applications, obstacle detection is a core capability for safe flight, in the meantime, it is a challenging problem in the real flight environment. This thesis presents an onboard obstacle detection system for the safe flight for a quadcopter. The system make use of a stereo vision for depth capture after an accurate calibration procedure. A long baseline and rigid mount of the stereo camera enable us to measure a long range obstacle from 0.5m to 15m. After getting the depth information, we use DBSCAN algorithm and some techniques to determine whether there is an obstacle in front of the quadcopter as well as the distance of the obstacle.