Use this URL to cite or link to this record in EThOS:
Title: Autonomous surveillance in an urban environment
Author: Oliver, John Douglas
ISNI:       0000 0004 2691 1785
Awarding Body: University of Warwick
Current Institution: University of Warwick
Date of Award: 2009
Availability of Full Text:
Access from EThOS:
Access from Institution:
A number of algorithms have been developed in the past for the purposes of target tracking, these have generally been for simple polygonal environments. However as the technology for autonomous vehicles develops for use in the real world these tracking algorithms need to be tested in larger more realistic environments. This work investigates the use of tracking algorithms to control a team of road based robotic platforms, tracking pedestrian targets in urban environments. The novelty of this work is in the identification of the aspects of the environment that affect target tracking algorithms, and modifying the algorithms to cope with them. Problems such as the frequent stalemates reached as an algorithms movement is limited by the highly restricted movement space or the identification of “short cuts” in which the target can take much shorter routes between positions than the robots. Algorithms are developed that overcome these limitations and they are tested in a simulation that is an accurate representation of a real environment. The algorithms are partly based on existing work and are developed extensively to be suitable for the environment. These algorithms are tested for their ability to maintain visual contact with the target. The scenario is tested with varying numbers of robots, speeds and locations. Three algorithms were developed and tested, one built as an extension of existing target tracking algorithms (Combined Urban Tracker) and another two algorithms developed specifically for this environment (Short Cut Path, and Branch). It is concluded that the Combined Urban Tracker and Short Cut Path algorithms performed comparably with a less than 0:3% difference in performance between the two both averaging roughly 54% effectiveness overall, however the Branch algorithm fared significantly worse averaging only 43% overall. The areas within the environment that give significant problems are large open spaces and areas that are significantly occluded from the road network. This work provides a platform on which further development in this area can be based in order to progress tracking algorithms towards being of practical use.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: TK Electrical engineering. Electronics Nuclear engineering