Anu Pradhan

Sensing and Field Data Capture for Construction and Facility Operations

Collection of accurate, complete, and reliable field data is not only essential for active management of construction projects involving various tasks, such as material tracking, progress monitoring, and quality assurance, but also for facility and infrastructure management during the service lives of facilities and infrastructure systems. Limitations of current manual data collection approaches in terms of speed, completeness, and accuracy render these approaches ineffective for decision support in highly dynamic environments, such as construction and facility operations. Hence, a need exists to leverage the advancements in automated field data capture technologies to support decisions during construction and facility operations. These technologies can be used not only for acquiring data about the various operations being carried out at construction and facility sites but also for gathering information about the context surrounding these operations and monitoring the workflow of activities during these o...

Investigation on Bridge Assessment Using Unmanned Aerial Systems

The U.S. currently spends tens of billions of dollars annually to inspect infrastructures and collect subjective, qualitative data that can often be unreliable or largely irrelevant. Inspections also require adequate access to remote locations, for example, appropriate scaffolding, lifting and additional protective equipment, which might increase the associated personal safety risk and add costs. The use of unmanned vehicles has experienced a tremendous growth primarily in military and homeland security applications. However, it is a matter of time until Unmanned Aerial Systems (UAS) will be widely accepted as platforms for implementing monitoring and inspection procedures. Researchers at Drexel University are exploring the use of quadcopters as vehicles to carry a set of remote sensors with the ultimate goal to perform bridge condition assessment. While the accuracy of remote sensing systems is somewhat limited compared to the one of contact sensing systems, the ability to quickly and periodically scan/inspect a structure without the need for scaffolding, ropes, or cherry pickers currently used during bridge inspections could transform the way the industry performs periodic bridge inspections. The Drexel team owns a number of UAS with different payload, flight time and range capabilities. In this paper, recent results obtained from preliminary testing on small mock-up concrete bridge decks as well as on small/medium size bridges are presented. One of the main efforts is to explore how multispectral imaging can provide a preliminary assessment of the deck condition of common highway bridges. Among future goals, Drexel’s team plans to develop and validate computer vision approaches leveraging data collected using UAS to permit geometric characterization (quantification of bearing position, girder deformations) and condition assessment (e.g. quantification of spalling and corrosion areas).

A taxonomy of reasoning mechanisms and data synchronization framework for road excavation productivity monitoring

Project management tasks, such as productivity monitoring and cost estimation, require data to be fused from multiple data sources, which are typically spatial and temporal in nature. In order to fuse a pair of spatial and temporal data sources, a number of different types of reasoning mechanisms are needed. This paper presents a taxonomy of spatial and temporal reasoning mechanisms needed to fuse spatial and temporal data sources to support construction productivity monitoring. In addition, the paper also describes two different approaches (i.e., interpolation and nearest neighbor approaches) that can be used to synchronize the temporal and/or spatial data sources. The developed taxonomy has been validated based on representative queries of construction engineers and managers that are identified in previous research studies. The interpolation and nearest neighbor approaches have been validated with real and simulated construction data sources.

Automated Detection of Damaged Areas after Hurricane Sandy using Aerial Color Images

Rapid detection of damaged buildings after natural disasters, such as earthquakes and hurricanes, is an urgent need for first response, rescue and recovery planning. In this context, post-event aerial images which could be collected right after disasters are valuable sources for damage detection. However, manual analysis process of the acquired imagery could be both time consuming and costly. To address this issue, a series of classification models for post-hurricane automated detection of damaged buildings is presented in this paper. First, five feature sets were generated through feature extraction and transformation. Then, several classifiers were trained using two groups of classification methods: (1) the Minimum-distance and (2) the Support Vector Machine (SVM) methods. The effectiveness of these classifiers was evaluated in terms of classification accuracies and testing time. The results demonstrated the combination of feature sets and classification methods can provide the best performance. Furthermore, optimal classifiers were selected for future automated real-time damaged building detection. The observed performances of these optimal classifiers indicate promising application for a wide variety of image-based classification tasks.

Formalized Approach for Accurate Geometry Capture through Laser Scanning

The main objective of the presented research is to formalize a novel iterative feedback mechanism that facilitates reliable and accurate geometric capture of civil infrastructure systems (e.g., bridges) through laser scanning to support predictive modeling (e.g., finite element modeling). Current condition assessment (of a bridge) based on visual inspection is subjective and often fails to obtain reliable and accurate geometry data which is essential for finite element modeling. The laser scanner, which is a non-contact geometry capture technology, is one of the promising technologies that can capture more detailed (2mm spacing) and accurate (+/- 5mm) geometric data compared to existing surveying and measurement techniques, such as total stations and Global Positioning Systems. The presented work is focused on investigating and formalizing how laser scanning technology can optimally be leveraged and integrated to provide a sufficiently accurate documentation of bridge geometry, and then how such information may be transformed into a reliable finite element model for predictive modeling.

Network-Wide Assessment of Transportation Systems Using an Epidemic Spreading Methodology

Network-wide assessment of transportation systems is a crucial task in traffic planning and management. Several simulation models (including macroscopic, mesoscopic, microscopic, as well as hybrid modeling) and theoretical models (e.g., macroscopic fundamental diagram) provide such assessments with different levels of detail. Although such models present good insight into dynamics of traffic flow (e.g., the relationship between speed and density) they do not discover the interrelationship between traffic flow dynamics and topology of the network. In this paper, the authors have attempted to macroscopically simulate congestion propagation on a regional road network based on the concept of epidemic spreading models (e.g., susceptible-infected-susceptible method). The results showed that at a critical threshold a transition occurs from a free flow phase to a congested phase. In addition, the simulation results verified that such a phase transition happens at the proposed mathematical threshold based on a topological characteristic (i.e., largest eigen value of adjacency matrix). However, the differences in the nature of congestion propagation and epidemic spreading suggest that further studies are needed to develop a hybrid method (i.e., macro-meso-micro) which mixes the proposed method with existing traffic simulation methods.

Binary and Multi-class Classification of fused LIDAR-Imagery Data using an Ensemble Method

Airborne Light Detection and Ranging (LIDAR) data is used for multiple applications, such as urban planning, emergency response, flood control, and city 3D reconstruction. The LIDAR data in its raw form needs to be classified for the above applications. There are two types of classifications: binary and multi-class. In the binary classification, the given LIDAR data is classified into two classes: terrain or non-terrain. In the multi-class classification, the given data is classified into multiple classes, such as ground, vegetation (low, medium, and high), and buildings. Although different techniques have been developed to address the challenges in LIDAR data applications in the last two decades, no single algorithm gives the best result. In this paper, we presented two Ensemble methods (Bagging and AdaBoost) that combine multiple algorithms in an intelligent way. These methods were developed and evaluated using 100 decision trees as the weak classifiers and combination of both point and neighborhood features. The authors were able to achieve the accuracy up to 98.9% for the binary classification and 94.6% for the multi-class classification. While AdaBoost performed slightly better that Bagging, the Bagging was more resistant to over-fitting.

Vulnerability of Infrastructure Systems: Macroscopic Analysis of Critical Disruptions on Road Networks

AbstractNetworked infrastructures serve as essential backbones of society. In particular, road transportation networks have a principal role in people’s everyday lives because they facilitate physical connectivity. External factors, such as disasters and failures, may degrade the performance of road networks. For example, severe flooding may disrupt a large area of a network, leading to cancellations and delay of several trips over the network. In addition, the collapse of a bridge could render certain nodes and links ineffective, thereby affecting traffic flow conditions. The authors formalized and developed a computational framework to assess vulnerability of road networks due to critical artificial or natural disruptions. The framework is built upon recent developments in interdisciplinary domains, such as network science, computational science, and transportation engineering. Using the proposed framework for the Greater Philadelphia road network, the network was divided into several high-critical, med...

Computational Modeling of Networked Infrastructures: Macroscopic Multivariate Approach

AbstractInfrastructure planners must perform many time-consuming simulations to prescreen numerous design alternatives. For example, transportation engineers must assess several alternatives to alleviate road transportation network congestion, such as constructing roads, changing organizations’ operating times, and changing shopping center locations. Evaluating alternatives requires considering changes in both structural and dynamic attributes of a network. A framework is proposed based on network science, computational science, and multivariate statistics to help planners train a model that can evaluate various alternatives efficiently. This model saves significant time and computational resources. The trained model takes structural and dynamic attributes as inputs and promptly returns macroscopic performance measures such as the averages and standard deviations of speed and volume over capacity. The model was trained using the Greater Philadelphia road network. Its average prediction error is 6%. Consid...

Automating and Optimizing Spatial Data Processing Workflows for Civil Infrastructure Inspection

Limited resources are available to timely inspect and maintain the aging civil infrastructure across the United States. Reality capturing technologies, such as laser scanning, is replacing visual inspection and manual surveying for improved data qualities and reduced resource requirements, while bringing challenges of timely processing terabytes of spatial data. Even using state-of-art 3D reverse engineering environments, inspectors need to manually select data processing algorithms, compose and configure data processing workflows, and verify the correctness of these workflows. Such manual design and execution of spatial data processing workflows are tedious, and result in sub-optimal workflows that do not fully utilize time and resources for producing accurate and detailed spatial information needed by domain applications. This paper proposes a computational framework that will assist in the infrastructure inspection process through streamlined spatial data processing workflow generation, execution, and optimization. Based on previous studies on spatial information query, spatial data processing, and building information modeling (BIM), the authors are exploring the feasibility of automatically generating and optimizing spatial data processing workflows based on formalized representations of these workflows.