Use this URL to cite or link to this record in EThOS:
Title: Obstacle detection in aerodrome areas through the use of computer vision
Author: Gauci, Jason
ISNI:       0000 0004 2686 5934
Awarding Body: Cranfield University
Current Institution: Cranfield University
Date of Award: 2010
Availability of Full Text:
Access from EThOS:
Access from Institution:
This thesis addresses the problem of ground collisions between an aircraft and obstacles (including other aircraft) on the ramp and taxiway regions of an aireld. A safety study is conducted by looking at current operating procedures and analysing accident statistics and reports. An onboard non-collaborative system for large transport aircraft is proposed and its main requirements and performance characteristics are discussed. The main requirement is to detect and track generic obstacles around an aircraft during taxi manoeuvres. The suitability of computer vision to the application of interest of this work is investigated through comparison with other candidate sensor technologies and computer vision, using visible cameras, is selected as the preferred technology. A study of dierent optical solutions is carried out and stereo vision is considered to be the most suitable choice. Two locations on the aircraft are considered for camera installation and the installation of a stereo vision system on each wingtip is chosen. Algorithms are implemented for the dierent processing blocks of the stereo vision system. These comprise calibration, rectication, correspondence, reconstruction, detection, clustering, and tracking algorithms. For each process, existing methods and techniques are reviewed and the most appropriate ones are selected, modied and improved in order to meet the specic requirements of this application. The values of several parameters of each algorithm are found experimentally using synthetic data and each algorithm is tested individually before being integrated with the rest of the system. Overall system performance is evaluated by testing for positional accuracy, generic obstacle detection and tracking capabilities, and sensitivity to calibration errors. Testing is conducted for a range of realistic conict scenarios, under dierent illumination, visibility, and image noise conditions. Both synthetic images and real images are used. The results of both sets of images are compared and these suggest that the stereo vision system developed in this research has the potential to reduce wingtip collisions and can therefore improve safety and situational awareness in aerodrome areas.
Supervisor: Zammit-Mangion, David Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available