Instant Segmentation and Fitting of Excavations in Subsurface Utility Engineering

Area of Science:

• Geomatics Engineering
• Civil Engineering
• Computer Vision

Background:

• Augmented reality (AR) in subsurface utility engineering (SUE) has advanced, but data quality remains a challenge.
• Existing SUE data often lacks completeness and accuracy, hindering effective utility management.
• Accurate as-built documentation is crucial for safe and efficient excavation and infrastructure projects.

Purpose of the Study:

• To present a novel automated approach for aligning existing SUE databases with real-time excavation measurements.
• To address the challenge of correcting deviations between as-planned and as-built utility documentation.
• To provide a solution that circumvents tedious postprocessing by correcting data online and on-site.

Main Methods:

• Infrastructure segmentation using live excavation site captures.
• A fitting approach to correlate inferred position and orientation with existing digital plans.
• Registration of as-planned models to the as-built state using automated data alignment.

Main Results:

• Demonstrated successful alignment of SUE databases with excavation measurements.
• Validated the proposed method's ability to correct as-planned to as-built documentation deviations.
• Achieved online, on-site data correction, eliminating the need for postprocessing.

Conclusions:

• The developed method offers a significant improvement in automating SUE data quality enhancement.
• This approach has the potential to revolutionize how as-built utility documentation is managed.
• Real-time data correction in the field enhances accuracy and efficiency for subsurface utility engineering.