Are pipe maps sufficiently accurate for augmented reality?

Subsurface utility excavation work planning would greatly benefit from Augmented Reality. Images like the one below sparked the imagination of many of our users, who say they could save a significant amount of time and guess work if they better knew the pipes location.

Of course, an augmented reality application does no pipe detection magic: it simply displays pre-existing pipe maps. Unfortunately, in most cases pipe maps only provide approximate 2D pipe locations. Inaccurate data would lead to inaccurate augmentations, which in turn might mislead the user. Such AR applications would be useless…

To provide good augmentation accuracy, an accurate pipe database is required. And such database should be updated regularly by surveying pipes every time they are installed or repaired. However, this would require a survey team available and ready at all time, which would likely increase the cost and duration of the excavation work.

Alternatively, one could also take photos of the pipes and their surroundings during the work, and use Bentley ContextCapture to produce a 3D mesh. The mesh would show the pipes with respect to buildings along with other more easily surveyable above-ground assets, and from which the absolute pipe location could be calculated.

All we would need is a way to capture those images autonomously and without delaying the excavation work. We need to attach cameras onto something that is already on site, and moving all the time.

For that purpose, we designed a simple rig, and attached 4 cameras onto it: 2 × Kodak SP360 4K, pointing sideways to provide context, and 2 × GoPro Hero5 Black, pointing down towards the excavation. We attached the rig onto an excavator boom, and used our cameras to capture videos of the excavation scene during the work.


We then synchronized the videos, extracted a representative 30-second section, and used Bentley ContextCapture to create a photo-realistic mesh. The resulting mesh clearly shows pipe location with respect to drains, houses, and other above-ground elements. And since the location of those assets can easily be measured, the true geo coordinates of the pipes can be inferred easily without delaying the excavation work.


Using such cameras mounted on excavators, the exact 3D location of pipes could be surveyed easily, at low cost, without delaying the excavation work. And over the years, the pipe database would become increasingly accurate, and progressively more useful for augmented reality excavation work planning. And the captured video files could even be stored for documenting the work that was done.

Perhaps such cameras will eventually become part of standard excavator equipment?