Rapid data quality oriented laser scan planning for dynamic construction environments

Abstract

In construction environments, laser-scanning technologies can perform rapid spatial data collection to monitor construction progress, control construction quality, and support decisions about how to streamline field activities. However, even experienced surveyors cannot guarantee comprehensive laser scanning data collection in the field due to its constantly changing environment, wherein a large number of objects are subject to different data-quality requirements. The current practice of manually planned laser scanning often produces data of insufficient coverage, accuracy, and details. While redundant data collection can improve data quality, this process can also be inefficient and time-consuming. There are many studies on automatic sensor planning methods for guided laser-scanning data collection in the literature. However, fewer studies exist on how to handle exponentially large search space of laser scan plans that consider data quality requirements, such as accuracy and levels of details (LOD). This paper presents a rapid laser scan planning method that overcomes the computational complexity of planning laser scans based on diverse data quality requirements in the field. The goal is to minimize data collection time, while ensuring that the data quality requirements of all objects are satisfied. An analytical sensor model of laser scanning is constructed to create a "divide-andconquer" strategy for rapid laser scan planning of dynamic environments wherein a graph is generated having specific data quality requirements (e.g., levels of accuracy and detail of certain objects) in terms of nodes and spatial relationships between these requirements as edges (e.g., distance, line-of-sight). A graph-coloring algorithm then decomposes the graph into sub-graphs and identifies "local" optimal laser scan plans of these sub-graphs. A solution aggregation algorithm then combines the local optimal plans to generate a plan for the entire site. Runtime analysis shows that the computation time of the proposed method does not increase exponentially with site size. Validation results of multiple case studies show that the proposed

As-is Geometric Data Collection and 3D Visualization through the Collaboration between UAV and UGV

“…It then selects sensing locations and plans the "minimum-time" path for moving the scanner to this location. Some studies have shown that the laser-scan planning problem can be solved using as-designed models [11], [12]. However, current practices still manually collect and process the 3D point clouds of large-scale environments, which requires many scans and often results in high labor and time costs, human errors, and inconsistent data quality.…”

UAV Mission Planning Using Swarm Intelligence and 4D BIMs in Support of Vision-Based Construction Progress Monitoring and As-Built Modeling

“…The viewpoint resampling was employed to design customized inspection routes that maximize the coverage and completeness of constructed 3D models for existing structures (Bircher et al 2016). While tailored to terrestrial laser scanning, scan stations were automatically identified while taking into account the accuracy and the level of detail of the resulting 3D point clouds of target structures (Zhang et al 2016). Another study further elaborated on the interoperable and IFC-based integration of UAV mission planning modules with progress tracking and automated as-built modeling solutions (Hamledari 2017)…”

“…To increase the accuracy of such vision-based techniques, it is crucial to investigate the design of UAV image acquisition strategies that ensure the high quality of captured data (Morgenthal and Hallermann 2014). The data captured during manually planned inspections suffers from detail, coverage, and accuracy (Zhang et al 2016); the construction dynamics and changes in the building layouts also increase the need for model-driven and automatic planning of UAV inspection missions.…”

From Semantic Segmentation to Semantic Registration: Derivative-Free Optimization–Based Approach for Automatic Generation of Semantically Rich As-Built Building Information Models from 3D Point Clouds

“…The latter, which can be collected using any type of 3D scanner, have become increasingly affordable and thus popular in recent years (Tang et al, 2010;Wang et al, 2017). Considerable advances have also been made in point cloud preparation and pre-processing proficiencies (e.g., Song et al, 2014;Zhang et al, 2016) in model generation methods (e.g., Bosché et al, 2015;Díaz-Vilariño et al, 2015) and for semantic enrichment (Belsky et al, 2016;Hamledari et al, 2017;Sacks et al, 2017). For example, by cloaking the geometry of an existing facility as meshes, a model can mimic the real-life object and its exterior and interior surfaces with millimeter accuracy.…”