E!114169 AGAVE: AI-based Geomodels from UAV Geoscanning for sustainable infrastructure development

AGAVE: AI-based Geo-models from UAV Geoscanning for sustainable infrastructure development The building and construction sector accounted for 36% of final energy use and 39% of energy and process-related CO2 emissions in 2018. Amongst the challenges in infrastructure projects is the limited knowledge about what is beneath the surface. When building roads, tunnels or bridges, it is essential to optimise resource utilisation in order to improve sustainability. It is estimated that globally, the lack of geological knowledge results in 90% of infrastructure projects encountering cost overruns of 20-50 %. Traditionally, costly direct ground investigations are used to investigate ground conditions. The conventional approach comprises of traditional methods, like geotechnical drilling and sampling. In contrast to those relatively slow and limited methods used to understand the ground conditions, airborne geoscanning efficiently illuminates the ground response to an electro-magnetic induction pulse generated by a large sensor. This technology reaches a penetration depth ranging from several tens to several hundreds of meters. However, such systems are currently so large and heavy that they require the use of helicopter in order to deploy, limiting their use to very large infrastructure projects. The goal of the AGAVE project has been to reduce the size and weight of such geophysical systems to allow for the use of lighter aerial platforms such as UAV (unmanned aerial vehicles) instead of costly helicopters, and to use machine-learning to merge the data with existing geological information in order to drastically improve the processing and interpretation of the remote sensing data to predict 3D models of the sub-surface. This will open up geoscanning technology to smaller projects, and allow the use of high resolution 3D models to reduce risk, and save both time and total cost by applying the new solution. In simple terms, we are developing a drone-based X-ray vision for investigating the underground. After three years of joint development work, the AGAVE team has proudly achieved significant milestones towards the project goal. Two key achievements are successful test operation of the drone based geoscanning solution and production ready cloud- and machine learning based data processing and model building technology. The approval of flying the drone system represents a significant achievement, marking, to our knowledge, the first of its kind in Europe. This accomplishment, spearheaded by project partners SDU and SkyTEM was corroborated during field tests in Denmark, where the data obtained from the drone displayed superior resolution when compared to existing data collected from the same location using ground based geophysical methods. System tests conducted in Norway and Denmark showcased how it is possible to present real world data with the current version of the instrument. This result was confirmed and stands as a remarkable achievement in the project. The team and new technology are now ready to deal with drone-generated data as the system comes to commercialization. Initial prototypes of workflows and algorithms were functional at the beginning of the project, and served as a foundation for initial geological prediction products. Throughout the project's duration, substantial progress was made on several fronts. New approaches to geological modelling were assessed, robust system architecture was developed, and advanced quality assurance tools, to name a few, were prototyped and implemented. Consequently, this new technology emerged as a more advanced and efficient alternative to traditional methods, offering significantly greater value to present and future customers. The current helicopter-based solution carries a substantial entry cost while offering comparatively lower resolution to drone based systems, limiting return on investment for smaller infrastructure projects reliant on precision and constrained ground investigation budgets. With the drone system already approved for flying in Denmark, the next phase after AGAVE involves transferring the approval to Norway over the next year in collaboration with project developers and contractors. Once fully developed, EMerald and SkyTEM aspires to elevate the UAV-based workflow and system to become the new standard for ground investigations, promising lower costs, faster deployment and superior accuracy and resolution. These developments increase the possible applications of such technology, contributing to a significant and historic push forward in the industry where safer, more comprehensive and lower-risk investments in infrastructure projects can be made, and sets the stage for exciting new developments in the future.

The desired impact of the project was to reduce the likelihood for budget overruns in infrastructure and greenfield development projects by cutting the total overall ground investigation costs. The effects of which are enabled through an integrated, high-resolution ground investigation workflow consisting of a geoscanning system carried by an unmanned aerial vehicle (UAV), machine- learning-based data integration and a drastically reduced invasive sampling program. The overall impact has successfully been demonstrated in recent geoscanning project carried out in partnership between EMerald and SkyTEM, albeit based on helicopter rather than UAV surveys. Recent railway planning projects in Brazil have demonstrated the project teams’ joint value proposition, demonstrated both risk reduction and a drastic reduction for the need of traditional ground investigation (drilling) programs. The developed ML workflows played, and will continue to play a key role in these achieved impacts and lay the foundation for future commercial deployments of the UAV sensor.

We will reduce geological risk for infrastructure development projects worldwide. We will develop a higher-resolution, UAV-based geoscanning system combined with machine-learning-based data processing workflows. Project owners and contractors will receive seamless 3D models with statistical precision over the full project area based on our data and a small number of strategically placed boreholes. Globally, 90 % of infrastructure projects encounter cost overruns of 20-50 %. This is costing society unacceptable amounts of time and money. A lack of thorough geological understanding is frequently identified as a key contributor to these overruns. Our technology will significantly improve knowledge and thus decrease this risk. Public funds can be saved, and decisions can be made with a better understanding of the involved risks, improving the sustainability of infrastructure development. Traditionally, costly direct ground investigations are used to investigate ground conditions. We have demonstrated that our method's return of investment is high when used to reduce traditional investigations. With our proposed development the price point will further decrease, due to the increased efficiency our margin will increase. With increasing market acceptance, our profits will benefit from an increased margin and market share.