Facial reconstruction

Search LJMU Research Online

Browse Repository | Browse E-Theses

A coordinated UAV deployment based on stereovision reconnaissance for low risk water assessment

Sam-Odusina, T (2018) A coordinated UAV deployment based on stereovision reconnaissance for low risk water assessment. Doctoral thesis, Liverpool John Moores University.

[img]
Preview
Text
2018temiphd.pdf - Published Version

Download (6MB) | Preview
[img] Text
2018temiphdinternal.pdf - Submitted Version
Restricted to Repository staff only

Download (6MB)

Abstract

Biologists and management authorities such as the World Health Organisation require monitoring of water pollution for adequate management of aquatic ecosystems. Current water sampling techniques based on human samplers are time consuming, slow and restrictive. This thesis takes advantage of the recent affordability and higher flexibility of Unmanned Aerial Vehicles (UAVs) to provide innovative solutions to the problem. The proposed solution involves having one UAV, “the leader”, equipped with sensors that are capable of accurately estimating the wave height in an aquatic environment, if the region identified by the leader is characterised as having a low wave height, the area is deemed suitable for landing. A second UAV, “the follower UAV”, equipped with a payload such as an Autonomous Underwater Vehicle (AUV) can proceed to the location identified by the leader, land and deploy the AUV into the water body for the purposes of water sampling. The thesis acknowledges there are two main challenges to overcome in order to develop the proposed framework. Firstly, developing a sensor to accurately measure the height of a wave and secondly, achieving cooperative control of two UAVs. Two identical cameras utilising a stereovision approach were developed for capturing three-dimensional information of the wave distribution in a non-invasive manner. As with most innovations, laboratory based testing was necessary before a full-scale implementation can be attempted. Preliminary results indicate that provided a suitable stereo matching algorithm is applied, one can generate a dense 3D reconstruction of the surface to allow estimation of the wave height parameters. Stereo measurements show good agreement with the results obtained from a wave probe in both the time and frequency domain. The mean absolute error for the average wave height and the significant wave height is less than 1cm from the acquired time series data set. A formation-flying algorithm was developed to allow cooperative control between two UAVs. Results show that the follower was able to successfully track the leader’s trajectory and in addition maintain the given separation distance from the leader to within 1m tolerance through the course of the experiments despite windy conditions, low sampling rate and poor accuracy of the GPS sensors. In the closing section of the thesis, near real-time dense 3D reconstruction and wave height estimation from the reconstructed 3D points is demonstrated for an aquatic body using the leader UAV. Results show that for a pair of images taken at a resolution of 320 by 240 pixels up to 21,000 3D points can be generated to provide a dense 3D reconstruction of the water surface within the field of view of the cameras.

Item Type: Thesis (Doctoral)
Uncontrolled Keywords: Stereovision; UAV
Subjects: T Technology > TA Engineering (General). Civil engineering (General)
Divisions: Engineering
Date Deposited: 05 Dec 2018 13:19
Last Modified: 18 Oct 2022 13:47
DOI or ID number: 10.24377/LJMU.t.00009766
Supervisors: Bezombes, F, Kot, P and Shaw, A
URI: https://researchonline.ljmu.ac.uk/id/eprint/9766
View Item View Item