Revolutionising Asset Management Through AI

Abstract

APP have partnered with Arrayen AI to investigate innovative ways to assist road authorities in the management of their road assets. The core foundation required in the management of assets is information. The traditional methods of logging defect data and subsequent analysis can be time consuming, resource intensive, and therefore costly. Without reasonable information it is difficult for anyone to make informed decisions about work priorities, budgets, medium- and long-term investment levels that are financially sustainable. Using Artificial Intelligence decreases the time required, reduces human resource inconsistency and the amount of effort which in turn decreases cost. The result is suitable information to drive informative decisions. How is this achieved? • Asset Capture – We have developed software that allows road asset information to be quickly and easily captured using cameras. This is one of the main solutions to gather data at normal traffic speeds. Once captured, the video data can be processed using specialised software developed by Arrayen AI. Within this process the captured data is combined with the data from road authorities Digital Road Network (DRN) to produce high-resolution images that align directly to road authorities existing chainage points. Specific chainage locations can be extracted from the captured video, meaning that the data can be consistently reproduced and ensures that it always aligns with road authorities DRN points. The images produced are also geo-tagged and have the chainage information, road name, and capture date/time stamped on them. We then take this a step further and run the captured data through our custom trained machine learning (ML) models to detect and identify different objects and situations within the images. Using our custom trained ML models, we can detect things like Traffic Signs, Road Damage (Potholes, Cracking), Pedestrians and specific types of vehicles (Cars, trucks, Buses etc). • Reports & Outputs - The information that is provided as an output, records defect type, location, and associated image with meta data. This information targets areas for further groundtruthing activities to the specific locations, to determine quantity of defect, and potential repair costs saving both time and cost. With AI technology it is possible to process 100,000 images in an hour which is 10km of the road network. For anyone who has ever undertaken a desktop review of imagery logging defects you will understand the significant time savings that AI can produce. The output information can be imbedded into client GIS systems, making it easier to identify hot spots, efficiently prioritise repair crews, determining future renewal programs and easier monitoring of progress. Applications – Dilapidation monitoring for Road User Agreements to inform the change in asset condition resulting from the proposed road use at set intervals. Informing asset registers with the compilation of street or road assets such as signs, seating, bus stops, hydrant makers etc. Maintenance backlog lists to inform maintenance works, operational budget expenditure, provide informed decision making on capital/renewal investment. Disaster Recovery Funding Arrangements Discuss future applications in the pipeline.