Digital twinning of asphalt pavement surfacings using visual simultaneous localization and mapping

摘要

Quantification of pavement surfacing characteristics relates performance indicators pertaining to the mean texture depth, coarseness and loss of aggregates. Traditional measurement methods require skilled personnel and methodical execution of established methods to indirectly infer these characteristics. High accuracy Visual Simultaneous Localization and Mapping (VSLAM) techniques has seen widespread adoption in areas of robotics and visual odometry. Advances in hardware miniaturization, digital processing of optical imagery and reconstruction has seen the development of commercial scanners that provide mobile platforms and accuracies comparable to that of traditional, laboratory-grade laser scanners. Incorporating state-of-the-art, non-destructive digitization technology, together with well-established investigative methods, aims to improve both the fidelity and speed of surface data acquisition. The pipeline for the creation of a digital twin is simple in its execution, producing sub-millimeter accuracy and dense reconstructions over a wide range of dimensional scales. The twin model enables not only calculation of texture and profile characteristics, but also describes the orientation and curvature of particles along with changes in surface coarseness over time. The digitization process ties in directly with the continued drive toward digitization as part of improving pavement rehabilitation using predictive analytics and machine learning, improving existing preventative maintenance strategies.