**1. Introduction**

The success of projects is evaluated through project completion within constraints of time, scope, cost, and quality. According to the KSA vision 2030 report, in 2017 alone, approximately 60% of construction projects were 20% behind schedule. In addition, more than 35% of the project time was spent collecting and analyzing data [1]. Further, approximately 15% of the construction cost was for rework activities. Therefore, time and cost were wasted in collecting and analyzing data, making as-built plans, monitoring the project, and fixing errors [1]. Therefore, researchers turned their attention to automated inspection to increase the response time of delayed activity rather than manual inspection [2,3]. Another example is that researchers were inclined to use automation methods to track and monitor the construction progress for better visualization after 2007 [4].

In this context, the construction progress monitoring domain (CPM) has developed massively in the last two decades. The exponential increase in computational capacities has allowed the architectural, engineering, and construction (AEC) industry to develop and implement automated methods in the CPM field. Lately, the development of the CPM field has depended on two primary methods: building information modeling (BIM) and 3D sensing technologies. BIM is focused on accurately establishing "as planned" 3D models. An as-planned model can also generate a spatial representation of project components.

**Citation:** ElQasaby, A.R.; Alqahtani, F.K.; Alheyf, M. Automated Schedule and Cost Control Using 3D Sensing Technologies. *Appl. Sci.* **2023**, *13*, 783. https://doi.org/10.3390/ app13020783

Academic Editor: Dimitris Mourtzis

Received: 29 November 2022 Revised: 25 December 2022 Accepted: 29 December 2022 Published: 5 January 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Then, integrating a 3D model with the project's information could produce an accurate 3D BIM-based model [5,6]. The four-dimensional model was also recognized as the schedule model (4D). The scheduling model has been designated to establish the activities' sequence over time. Cost information on the project's activities is another dimension of BIM known as 5D BIM. Activities' completion and cost over time have been simulated in a virtual environment. There have been limitations to the current BIM-based cost model, such as the cash flow analysis [7]. Some researchers also identified the sixth dimension as the facility phase. However, other studies referred to the sixth dimension as sustainability and its implementation in smart cities [8].

#### **2. Previous Studies**

3D sensing technologies are another crucial aspect that improved immensely track and monitor the progress of construction components. These technologies included radio frequency identification (RFID), an ultra-wideband system (UWB), a global positioning system (GPS), image processing methods, and laser scanners (LS) [9]. Previously, researchers managed to assemble "as-built" models [10], where a developed as-built model was created to restore, record, and improve historic buildings. Another study investigated integrating BIM and remote sensing instruments where a BIM-based model with a laser scanner was integrated for quality control in real-time to reduce schedule and cost overrun [11].

Among the 3D sensing technologies, A laser scanner, also known as Light Detection and Ranging (LiDAR), is one of the AEC industry's most recognized technology. Laser scanning aims to map 3D objects into point cloud datasets [12]. Similarly, researchers monitored and controlled the infrastructure components using as-built data using a laser scanner [13]. Laser-based methods have also been used in recognizing construction applications such as workspace modeling, asset management, and worker tracking [14]. Another laser scanner application tracks buildings' temporary or secondary components [15]. Although there are other 3D sensing technologies, laser scanners (LS) are one of the best-fitted technologies to track and monitor the 3D status of projects accurately [16–19]. In addition, researchers used automated methods as they have lesser limitations and could save much work and time in assessing the progress of construction projects [16].

3D spatial technologies were used to monitor and control the progress in the CPM domain. Some researchers applied the RFID system to form an as-built model, while others used UWB systems [20,21]. The image processing technology was also used in the CPM field using digital images or UAVs of construction activities [22,23]. Point cloud data sets were similarly used to evaluate the progress in construction buildings through laser scanning technology [19,24]. However, researchers used more than one sensing technology for more robustness and better results; for example, a study conducted by [25] used more than one 3D sensing technology (UWB system and laser scanner). Another study used a combination of RFID and laser scanning technology [26]. Other studies have used a fusion of image processing methods and laser scanners [24,27–29].

Furthermore, some latest review articles discussed different insights to recognize knowledge gaps and recommend future directions in the CPM field. For example, the methodology in [4] applied scientometric analysis to point out a broad picture of CPM. Another example is a systematic literature survey conducted to automate indoor progress monitoring [30]. The 3D model reconstruction and geometry quality inspection were also discussed comprehensively using the point cloud datasets [31]. Meta-analysis was estimated to review the quality of studies of object recognition performance indicators in the CPM field between 2007 and 2021 [32]. Previous studies recommended the usage of 5D assessment for future research to address the gap in the BIM integrated with 3D sensing technologies in CPM applications.

Therefore, the contribution of this paper is to propose an automated construction progress tracking system for schedule and cost control. The reason behind that is the lack of research on conducting 5D assessment in the CPM domain using EV principles [4,19,30–32]. The proposed system automatically implements a 5D assessment: progress feedback regarding schedule and cost per scan. The 5D assessment enables reviewing the progress and states the project condition through EV principles (schedule performance index, cost performance index). The proposed system outcomes were also compared to the manual system to validate the accuracy of the proposed system.

#### **3. Methodology**

This paper illustrates an automated progress tracking system in construction progress monitoring to assess the updated information in schedule and cost. A BIM-planned model was established from 2D shop drawings. Then, the as-built model was established by collecting scans using laser scanning technology, processing them, and registering them to a common coordinate system. The as-built model would also be evaluated and assessed to determine the quality of point cloud sets based on three KNN searching algorithms: a fixed number of nearest neighbors, a fixed neighborhood radius, and an adaptive neighborhood radius. Once the integration between those models was automatically established, two algorithms were designed and developed to recognize the as-built objects. Another two algorithms were developed to review the progress in terms of schedule and cost (4D, 5D) using EV principles. The flowchart of the proposed approach is shown in Figure 1. Further explanation of the methodological steps is provided in the following sections.

**Figure 1.** Steps of Methodology.

#### *3.1. Tools*

In order to apply the methodology mentioned above, a set-up of a BIM-planned model and an as-built model was crucial to be established. Revit interface was used to establish the BIM-planned model by converting 2D shop drawings to 3D models. Then, the planned schedule and cost models were manually established based on material, equipment, and labor costs for each project milestone. Material costs include supplies or materials purchased for the project, such as concrete, walls, and rebars. The transportation and storage cost is also included in the cost of materials.

The data acquisition was conducted on-site using a laser scanner Faro Focus3D because laser scanners are mainly accurate and efficient [33,34] (See Section 3.3). Then, datasets were processed, the scattered points were transformed into a range image, and laser scans were registered using project reference points from one of the local coordinate systems of multiple scans to a common coordinate system [35,36]. Iteratively closest point (ICP) was then used in registration where correspondences between points of a scan called the source and points of another called the target was established to minimize the spatial distance between points in each pair. The reason behind using ICP was to achieve satisfactory registration results [37]. The outcome of the previous procedures was to generate the as-built model.

Once the as-built model was established, the next step was to indicate the strength or the weakness of the spatial distribution of the datasets by feeding the as-built model with K-nearest neighbors search algorithms (fixed number of nearest neighbors (Method I), fixed neighborhood radius (Method II), and adaptive neighborhood radius (Method III)) [38,39]. One million points were used as a reasonable sample because using the original point clouds is not computationally feasible. Firstly, the KNN algorithm based on a fixed number of neighboring points was set as [500, 5000] with an interval of 50. Secondly, the KNN algorithm based on a fixed neighborhood radius was explored using a neighborhood threshold between 5 cm–50 cm with a step length of 5 cm. The neighborhood threshold considered the further analysis of geometric features of columns, beams, and slabs when setting the threshold radius. Finally, the KNN method based on adaptive radius was set between 1–30 cm with an interval of 1cm by calculating the information entropy of the neighboring point cloud set [40]. The chosen lower band was set based on the point cloud noise, density, sensor specification, and computational constraints. However, the chosen upper band was set based on the most significant object in the scene (facades for LS-data sets) [41]. The geometric features shown in Table 1 were obtained to illustrate the spatial distribution of the datasets based on the KNN searching algorithms [41–43].


