Investigating the accuracy and reliability of virtual reality and UAV photogrammetry models in engineering design

Abstract

Throughout the last 30 years, information technology has experienced an extraordinary growth and the world has turned into a digital platform where many basic functions of everyday life have been digitised and are now automated. The construction and engineering industry has been at the forefront of this digital age and have continuously harnessed new technologies to their advantage for increased design support, better communication, and greater profit margins. Two technologies, virtual reality and photogrammetry have been established within the industry with virtual reality (VR) used as powerful visualisation tool for greater understanding of a projects spatial layout and photogrammetry being used to obtain working three dimensional (3D) models of sites. There is, however, a lack of understanding of if these digital technologies can accurately present a real-life environment, if the data obtained from them is accurate and if they can be used as a reliable design tool within the industry.

Previous research into the accuracy of virtual reality models has highlighted that the simulated virtual models can be useful for project analysis, increased understanding and as a communication tool, but there is a lack of information on the benefits to a design team, for use in identifying design flaws or comparable to the traditional method of viewing 2D drawings. Photogrammetry has been well researched and shown to be accurate, but not in direct comparison to equipment that is already embedded within the industry such as laser scanning. To determine if these technologies are accurate and reliable, two experiments were undertaken to analyse and evaluate both technologies. A user focused experiment was carried out to test a participant’s ability to detect a number of design flaws within a custom-built 3D design using both viewing methods; 2D and 3D VR, as well as extracting data on a user’s understanding of shape, layout and structure. The second experiment analyses the difference in dimensional data gathered using UAV photogrammetry methods and laser scanning to determine if photogrammetry is as accurate as laser scanning and can be used as a reliable surveying source. Results shows that participants could accurately identify 8% more design errors within VR than within 2D design drawings and that they could correctly estimate dimensions within VR to within 0.5m. It also showed a 57% increase in the understanding of structure. It was shown that data gathered by the UAV to create 3D photogrammetry models was accurate within specific survey accuracy bands as well as, on average, costing a third less than a laser scan survey. While it is not as accurate as LiDAR, outcomes show that it can be used as a potential survey tool within the engineering design process. The results shows that both technologies provided beneficial, reliable, and accurate data visualisations of their respective 3D environments.

Date of Award	Jul 2024
Original language	English
Awarding Institution	Queen's University Belfast
Sponsors	Engineering and Physical Sciences Research Council
Supervisor	Daniel McPolin (Supervisor), Sree Nanukuttan (Supervisor) & Karen Rafferty (Supervisor)