Based on a report from DIRT, Common Ground Alliance (Reference) in 2021, 228,393 damages happened in US and Canada. According to equipment world (Reference) one of the main reasons for most of these damages are caused by “Utilities not marked or marked inaccurately due to locator error”. The vulnerability of underground assets and infrastructures during excavation poses a significant obstacle. Resolving these issues is indispensable for streamlining construction workflows, ensuring safety, and optimizing project outcomes. As construction equipment and machinery operate in close proximity to these utilities’ underground assets, there is a substantial risk of accidental damage. Traditional methods of utility detection may fall short in accurately pinpointing these critical elements hidden beneath the ground. This poses not only financial risks due to potential project delays and repairs, but also threatens worker safety and the integrity of urban infrastructure.
One of the potential solutions to tackle this issue is by using Mixed Reality (MR) applications, however, there are a couple of issues with initializing the position of the users in urban areas. GPS positioning in urban environments, characterized by high-rise buildings, leads to inaccuracies and unreliability of GPS signals. These inaccuracies are amplified in the context of MR applications, where precise spatial positioning is crucial for overlaying digital information onto the physical world. Our proposed application solves this issue by taking advantage of machine learning in computer vision to find matching points by having an image dataset. Moreover, the results are synced with our GIS scenes.
By providing construction teams with accurate, real-time information about the locations of underground utilities and infrastructure, our solution aims to prevent accidental damages, ensuring both the safety of workers and the integrity of critical infrastructure components.
Our solution deploys a multifaceted approach, harnessing live-camera video and Google Street View to identify matching points. By skillfully intersecting the lines of these matching points, a substantial boost in GPS accuracy is achieved. Developed using ESRI SDK and ARCore, our application boasts two distinct modes tailored to distinct scenarios. The first mode delivers a world-scaled perspective for remote model interaction, supplemented by the utilization of Gyroscope and Accelerometers. The second mode seamlessly integrates mixed reality with our GIS web scenes, guaranteeing real-time updates.
The developed solution aids construction management by supplementing underground infrastructure documentation with an accurate 2D/3D virtual representation. This visualization helps construction teams identify potential challenges, plan logistics, and streamline workflows before starting on-site activities.
