Civil Engineering September 2021 | Vol 29 No 8

30 September 2021 Civil Engineering Conclusions The initial proof of concept study shows that bridge inspections are possible using drones and photogrammetry. Defects are clearly visible from the photos if sufficient natural illumination is present. The point cloud model adds context to the defects as a fourth dimension to the DER rating methodology. Investigating the defects on the point cloud model gives perspective on where the defect is located and provides more information regarding the relevancy of the defect. Improved photogrammetry software can potentially increase the quality of the point cloud model and could enable inspectors to identify additional defects that would normally not be identified when doing a regular visual assessment. Using higher quality cameras will make it possible to capture images underneath structures where the natural light is low. The use of drones for network level inspections, as required by provincial and municipal road authorities, will be a challenging task as permission would be required to capture images using drones for a large number of structures, some of which are located on high volume roads, close to airports or in residential areas. A methodology for network inspections is needed to ensure consistent data cap- turing and optimised use of technology (cost versus sufficient quality). Bridge and senior bridge inspec- tors need to be highly qualified and experienced. Using such individuals to inspect all the bridges and culverts belonging to a road authority is a time consuming and costly exercise. If point cloud models could be used to reduce the number of structures a bridge inspector has to physically inspect on site, the cost of network inspections could be significantly reduced. It is important to note that drones are only the enabler and that the real value lies in the images that are captured and processed. Drones would be useful when inspecting high structures or those over flowing rivers where it would be difficult for an inspector to reach the entirety of the structure. For lower bridges and culverts with confined spaces it is not practical to use drones. For such structures, hand gimbals could be used to capture the images which can then be processed using photogrammetry to create the same point cloud models. Future Work The focus of the initial proof of concept for using 4IR technologies to improve bridge inspections in South Africa was on the pos- sibility of identifying bridge defects using drones and processed images. The two bridges that formed part of this study were visually assessed in 2016. The defects on these bridges and the location of these de- fects were therefore quantified beforehand. During the next phase of the study, im- ages of different bridge structures will be captured using drones and hand gimbals. The point cloud models will then be given to an independent, COTO-accredited bridge inspector and he/she will attempt to identify defects and provide DER ratings using only the point cloud models. The defects identified and the DER ratings allo- cated to the defects will then be compared with those from previous TMH 19 visual inspections on site. This blind comparison will indicate if it is possible for a bridge inspector to identify and rate defects using point cloud models only, without their presence on site. It will also indicate if additional defects and different types of defects can be identified using the point cloud models. Lastly, it will provide an indication of the confidence level of the inspector using point cloud models exclusively. The following aspects will be included for further development in future studies: Q Q Methodology for network inspections Q Q Different imaging technologies e.g. thermal cameras Q Q Inspection of different types of structures Q Q The possibility of using the existing database (inspection photos) to assign defects and apply machine learning techniques Q Q Integration of point cloud into a BMS Q Q Adjust the layout of inspection sheets and BMS user interfaces to accommo- date the image processing inspection methodology.  REFERENCES Ciampa, E., De Vito, L., & Pecce, M. R. 2019. Critical issues on the use of drones for constrution inspections. Journal of Physics: Conference Series. IOP Publishing. COTO. 2017. TMH 19 Manual for the visual assessment of road structures. Pretoria: The South African National Roads Agency SOC Limited. Kock, S. 2015. An overview of South African RPAS regulation. Geomatics Indaba Proceedings. Ekurhuleni. SACAA. 2017. Remote Piloted Aircraft Systems. Retrieved from http://www. caa.co.za/Pages/RPAS/Remotely%20 Piloted%20Aircraft%20Systems.aspx# Wells, J. & Lovelace, B. 2018. Improving the Quality of Bridge Inspections Using Unmanned Aircraft Systems (UAS) . Minnesota: Minnesota Department of Transport. DIVERSE RANGE, UNDENIABLE QUALITY! BECAUSE YOUR JOBSITE, DESERVES THE BEST. All products and services on www.wackerneuson.co.za