Construction Research Congress 2012 · 2012 · 38 citations · 14 references
Construction JobsitesEngineeringLaser Scan PlanningField RoboticsPoint Cloud ProcessingComputer-aided DesignAutomated Scan PlanningSocial SciencesConstruction AutomationData Collection TimeData ScienceCalibrationSystems EngineeringLaser-based SensorAutomation In ConstructionComputational GeometryGeometric ModelingCartographyMachine VisionSurveyingSpatial Data AcquisitionSensor ModelingComputer ScienceData Collection DistancesConstruction OperationsAutomated InspectionCivil EngineeringAutomationConstruction Management3D ScanningConstruction Engineering
Terrestrial laser scanners on construction sites require careful planning of scan locations and technical settings, but time constraints, occlusions, and spatially varying information needs make manual design difficult, often resulting in longer data collection, interference, and missing data. The paper proposes an automated laser scan planning approach that uses sensor models of terrestrial laser scanners. A sensor model linking collection parameters to point‑cloud accuracy and density is combined with an optimization algorithm that selects scanning plans minimizing data‑collection time while meeting user‑specified quality and coverage requirements.
A challenge of using terrestrial laser scanners on jobsites is to effectively plan the scanning locations and configure scanners' technical parameters according to the domain requirements and site conditions. With time limits in the field, it is challenging to manually design and analyze scanning plans considering technical capabilities of scanners and environmental conditions (e.g., occlusions) for ensuring that all needed spatial information is captured with the required levels of accuracy and detail. In addition, the information requirements of construction tasks may vary spatially and temporally, further complicating the scan planning. These facts can lead to longer data collection time, unexpected interferences on jobsites, and missing data. This paper proposes an approach of automated laser scan planning based on sensor models of terrestrial laser scanners. We developed a sensor model capturing the relationships between various data collection parameters (e.g., data collection distances) and data quality metrics (e.g., accuracies and densities of 3D point clouds). Given the user-specified data quality requirements, an optimization algorithm can use this sensor model to identify scanning plans that minimize the data collection time while optimizing the data qualities and spatial information coverage.
14