Integrating mobile Building Information Modelling and Augmented Reality systems: An experimental study

Automation in Construction 85 (2018) 305–316

Contents lists available at ScienceDirect

Automation in Construction journal homepage: www.elsevier.com/locate/autcon

Integrating mobile Building Information Modelling and Augmented Reality systems: An experimental study

MARK

Michael Chua, Jane Matthewsa, Peter E.D. Loveb,⁎ a b

VDC Cooperative, Department of Construction Management, Curtin University, GPO Box U1987, Perth, Western Australia 6845, Australia Department of Civil Engineering, Curtin University, GPO Box U1987, Perth, Western Australia 6845, Australia

A R T I C L E I N F O

A B S T R A C T

Keywords: AR BIM Cloud-based Experiment Design science Task efficiency

The benefits of Building Information Modelling (BIM) have typically been tied to its capability to support information structuring and exchange through the centralization of information. Its increasing adoption and the associated ease of data acquisition has created information intensive work environments, which can result in information overload and thus negatively impact workers task efficiency during construction. Augmented Reality (AR) has been proposed as a mechanism to enhance the process of information extraction from building information models to improve the efficiency and effectiveness of workers' tasks. Yet, there is limited research that has evaluated the effectiveness and usability of AR in this domain. This research aims to address this gap and evaluate the effectiveness of BIM and AR system integration to enhance task efficiency through improving the information retrieval process during construction. To achieve this, a design science research approach was adopted that enabled the development and performance of a mobile BIM AR system (artefact) with cloud-based storage capabilities to be tested and evaluated using a portable desktop experiment. A total of 20 participants compared existing manual information retrieval methods (control group), with information retrieval through the artefact (non-control group). The results revealed that the participants using the artefact were approximately 50% faster in completing their experiment tasks, and committed less errors, when compared to the control group. This research demonstrates that a minor modification to existing information formats (2D plans) with the inclusion of Quick Response markers can significantly improve the information retrieval process and that BIM and AR integration has the potential to enhance task efficiency.

1. Introduction Low levels of task efficiency have been and remain a pervasive problem for the Australian construction industry. Instead of focusing on people, increasing emphasis has been placed on the use and development of technical-based solutions associated with information and communication technology (ICT), specifically Building Information Modelling (BIM) to deliver ‘value’ in construction (e.g., [1,2]). However, there is limited empirical evidence to substantiate claims that the use of BIM leads to increases in task efficiency; though, it should be acknowledged that pockets of research have demonstrated workflow improvements for precast and modular construction, scaffolding erection and safety within a BIM environment (e.g., [3,4]). If BIM is to deliver ‘value’ then structures and processes need to be re-engineered to accommodate new workflows that are engendered by implementing BIM [5–7]. For asset owners who are dependent on the use of a building information model for operations and maintenance their ‘value proposition’ will need to be amended to adapt to the changes that will be ⁎

required to their intra and inter-organizational business processes [5,6]. While technology enabled systems such as BIM have their merits, the pace at which they are evolving and their capability to capture significant quantities of data, raises concerns as workers are confronted with too many information systems [8]. This can result in information overload, and negatively impact on task efficiency, with workers spending increasing amounts of time managing data on complex systems, rather than gaining the benefits [9,10]. Consequently, for systems such as BIM, that are dependent on the transfer of information through models, utilization is affected, as workers perceive less value in the system due to the increased effort required to retrieve information [11]. To improve the access and utilization of information, Augmented Reality (AR) has been identified as a technology that can be used to enhance the process of information extraction from building information models [12,13]. AR is an enhanced version of reality created by the use of technology to overlay digital information of an artefact when viewed through a device such as tablet or smartphone camera [14]. However, research demonstrating AR effectiveness or usability from a

Corresponding author. E-mail address: [email protected] (P.E.D. Love).

http://dx.doi.org/10.1016/j.autcon.2017.10.032 Received 15 February 2017; Received in revised form 12 September 2017; Accepted 28 October 2017 0926-5805/ © 2017 Published by Elsevier B.V.

Automation in Construction 85 (2018) 305–316

M. Chu et al.

automated, resulting in improved workflow performance [10], BIM enabled systems can complicate and overload processes with non-critical project information. This can have a negative impact on worker's performance and productivity, requiring them to spend more time manipulating or managing the data on complex systems, rather than reaping its potential benefits ([9,10]). Consequently, this can lead to a ‘productivity paradox’ being experienced [5,24]. For investments in BIM to be justified, optimal productivity of workers engaging with the technology must be attained with systems that are utilized by all stakeholders [10,25]. Existing barriers to collaboration and interoperability, need to be overcome, with an emphasis being placed on improving the ability to extract information effectively and efficiently from a building information model [26–28].

scientific perspective remains scant [14,15]. In addition, there is a distinct lack of developments that demonstrate ‘proof-of-benefit’ to construction tasks [13,15]. This is further compounded by difficulties associated with developing AR systems, as they tend to exist on nonstandardized hardware and software, which has hindered their adoption by the construction industry [15]. According to Wang et al. [15] AR and BIM are complementary technologies, but their integration has tended to be examined and evaluated from a technical standpoint rather than examining their potential to enhance the performance of work tasks. Researchers and practitioners within the construction industry rarely possess the knowledge base required to solve the prevailing issues that exist to effectively utilize AR technologies (e.g. software and hardware development). Consequently, Meža et al. [13] has suggested AR applications from disciplines such as gaming, media and marketing should be used as a source of innovation for construction. Against this contextual backdrop, the research presented in this paper utilizes a design science research approach to evaluate the effectiveness of BIM and AR system integration that is designed and developed to enhance efficiency through improving the information retrieval process during construction. By adapting existing AR development kits/tools that would typically be used for non-construction purposes, this research demonstrates the appropriateness of using pre-existing functionalities from commercially available AR software and hardware. In addition, the benefits for future researchers of leveraging performance capabilities from existing defined software applications are identified.

2.1. BIM utilization The benefits of BIM have typically been tied to its capability to support object oriented information structuring and exchange through the centralization of information [18]. It has also been acknowledged that BIM has the potential to deliver benefits for facility and asset managers, who have traditionally been left with incomplete or obsolete documentation to support their activities, during operations and maintenance (e.g., [6,21]). Research has demonstrated that changes made during construction are rarely federated into a building information model for use later by construction and operations teams [11]. This results in a series of sub-models which do not represent what has actually been constructed, which stymies an asset manager's ability to effectively use them during operations and maintenance. To develop an accurate federated model for handover to the operations team there needs to be collaboration between disciplines. Yet, Fazli et al. [26] revealed that organizational difficulties exist when justifying the initial cost and time requirements for training staff to work efficiently with BIM throughout a project's life-cycle. This has contributed to designers tending to perceive less value (i.e. gain/return on time and costs spent) in generating fully interoperable, federated models. This has been reinforced by Kerosuo et al. [11], who found that designers tended not to regard fully integrated models useful for their activities and often avoided the collaborative co-designing features of software. This has typically restricted the use of BIM to the design stages [20]. Even when models are transferred from the design to construction stage, site teams have tended to not to find them useful, instead preferring to use traditional 2D paper plans [11]. A shift from conventional standalone frameworks to cloud-based BIM systems to support full collaboration between stakeholders has been initiated [7]. While cloud-based BIM systems have paved the way for future offerings to support mobile workers access to a centralized database, they still tend be archival in nature and have yet to fully address issues associated with visualization [29]. To improve the effectiveness of BIM across an asset's life-cycle, integration with AR is required to enhance the extraction of building and product data that has been captured [13]. This creates a world whereby an AR user can interact with real-time augmentations of virtual information in their current environment that is contextual in nature and therefore enable improved decision-making [30,31].

2. Digitization in construction The last decade has seen significant improvements in the digitization (i.e. process of converting information into a digital format) of information in construction. The increased capabilities of new ICT systems to capture and manage information in projects, have, over time, created information intensive work environments. This has enabled construction personnel to gain near on-demand access to project data, plans, drawings, schedules and budgets [16]. Increased mobility due to technologies such as laptops, tablets and smart-phones, provides access to relevant and up-to-date digital building data whenever and wherever needed. As a result of BIM and the emergence of smart technologies, the construction industry is at the cusp of a transformative change [17]. To leverage the benefits of technology, such as BIM and AR however, an improved understanding of the information required by workers to carry out construction-related tasks is needed. A BIM environment is supported by an array of interrelated databases in a central location, which can accrue and store data over time [18]. This centralized resource encourages collaboration between stakeholders, allowing them to exchange information, query, simulate and estimate activities throughout a project's life-cycle [19]. More often than not, however, BIM is simply perceived as a tool for visualizing design and coordinating construction activities on-site [20]. It has been suggested that the integration of BIM with on-site construction processes can provide an array of benefits such as: [8] improved coordination; [9] clarity in task requirements; and [10] a reduction in misunderstanding of project requirements [20,21]. This in turn translates to eliminating waste, minimizing transaction costs, as well as enhancing the transfer of shared knowledge and expertise among all parties in a project [22,23]. Taking advantage of the recent growth in cloud-computing capabilities, the construction industry has begun to shift toward the adoption of cloud-based BIM solutions such as Autodesk's BIM 360™ Glue® and Viewpoint™. The resulting ability to access information ‘ondemand’ has provided a platform for the digitization of information during the construction process. While there are many positive aspects to using cloud-based solutions, there is also concern that workers are being provided with by too many systems to effectively communicate and collaborate [8]. While many non-value adding activities can be

2.2. Augmented reality Information-intensive activities that rely on paper mediums for information retrieval such as those undertaken during construction are well suited to AR [32]. The contextual awareness of AR systems enhances the process of information retrieval by providing a mechanism to filter data, information and services, thereby removing redundant data and allowing the user to see only relevant information [16]. The feasibility of AR in augmenting construction related-tasks can be assessed using the following cognitive principles ([32]; p.316): [8] information searching and accessing; [9] attention allocation [10] 306

Automation in Construction 85 (2018) 305–316

M. Chu et al.

By utilizing and adapting AR development kits/tools that would typically be used for non-construction purposes, the research therefore demonstrated the benefits of leveraging performance capabilities from existing defined AR software applications, which can be adopted in future studies [13]. In doing so, this removes the need to develop the core functionality of the software/hardware to enable the use of AR in construction. This allows a greater focus on the evaluation of the system to address problems that industry practitioners are confronted with in construction practice. This research, therefore, presents an alternative approach to previous AR studies, by developing a mobile cloud-based BIM AR system using a scientific approach. The system utilizes commercially available software and hardware and appropriates their preexisting functionalities. Three criteria were used to select the AR software to be used in the proposed system: [8] scalable for use on a nonisolated standardized system; [9] readily adapted to visualize an object in a building information model and enable extraction of information; and [10] able to link to a database that could contain primary data related to an object in a building information model to support the mobility of workers.

working memory; and [19] spatial cognition. Building upon this research, Hou et al. [12] and Nakanishi and Sato [33] have suggested that through a filtering mechanism a user may also benefit from improved cognitive function, which can enable workers to more effectively undertake assembly tasks by reducing cognitive loading through improving working memory capacity. In this instance, a user's ability to interpret and retrieve information is also significantly improved. Despite the potential of AR, its applicability and usability in construction remains limited due to issues surrounding its portability, functionality and its capabilities to transfer information to workers on their own exclusive platforms [14,15,34].

3. Research approach The integration of BIM and AR systems has generally been evaluated from a technical performance perspective using bespoke platforms (e.g., [18,29,31]). Such platforms are not readily available and are cumbersome to use in practice. Akin to Meža et al. [13], who relied on existing AR software and amended it for the purpose of construction, this paper aims to integrate BIM and AR technologies to enhance construction workers ability to retrieve information and thus improve their productivity. To achieve this aim a design science research approach is adapted [35]. The objective of this approach is to produce a BIM AR system that can be utilized by other researchers, by defining nominal processes to support further development/enhancement of existing knowledge and to facilitate the creation of new concepts [35,36]. Design science differs to other explanatory approaches, which tend to focus on describing, explaining and predicting the current natural or social world, by not only understanding problems, but also designing solutions to improve human performance [37]. By initiating a methodical process to acquire research knowledge and understanding of the problem area, artefacts are designed and applied, and can be further developed. Given the aim of this research, design science is suitable approach to adapt. The research process that was undertaken is summarized in Fig. 1.

3.1. Design development of BIM AR system The design development of the BIM AR system commenced after a review of the extant literature and acquired an understanding of the information retrieval process, denoted as steps 1 and 2 in Fig. 1. It was revealed that Wikitude provided a robust platform for AR development as all of its developmental features were available through an open source Software Development Kit (SDK) that allowed for full customization of its source code. Thus, this enabled its advanced features such as geo-based tracking, cloud recognition, and radar, to be developed for an Android application. Alternatively, developers may also use the web based user portal to create simpler augmented experiences without the need to programme code. This is available through the Wikitude's Studio features, which allows for the assignment of a 2D marker. In this research, the Wikitude New Studio (Beta) was used to create

Fig. 1. Design science approach: research process.

307

Automation in Construction 85 (2018) 305–316

M. Chu et al.

Fig. 2. Wikitude New Studio (Beta) user interface.

Fig. 3. Conversion of BIM Object to Wikitude AR format workflow.

specific information in the form of text and graphics, when viewed through the smartphone camera (Fig. 4). The database is accessed through the Wikitude mobile application

the mobile augmentations for the evaluation (Fig. 2). It provides the same features as the normal studio version, but with an enhanced user interface. Augmentation options include text, images, buttons, html links, 3D models or videos. Once created and uploaded to its cloud database, the augmentation is instantly accessible and can be visualized through the camera of any supported device. The developers at Wikitude have also made available a user-friendly tool for the conversion of the 3D building object models to its proprietary WT3 format. The workflow for conversion for generating BIM objects for AR process is identified in Fig. 3. Notably, the Augment AR software for Android platforms which is available through the Google Play Store® was also considered. And is available through the Google Play Store®. Unlike the Wikitude platform it focuses on the commercial consumer, acting primarily as an interfacing tool for media, marketing and advertising purposes. The software is proprietary and does not allow for the same level of development as Wikitude, though it does allow for the assignment of 3D building objects to markers. Therefore, the Augment platform was not used, as it could not visualize additional layers of information in a manageable way. Wang et al. [15] and Wang et al. [14] provide a comprehensive list and features for other toolkits for AR prototyping. Following the conversion of the 3D BIM object models for AR visualization, the BIM AR system was developed for operation on a Samsung Galaxy® S6 mobile smartphone, which operates an Android operating system. The Wikitude New Studio (Beta) platform provides a cloud database for storage of 2D markers or ‘Targets’. The target images used in this research were randomly generated Quick Response (QR) markers by goQR [38]. The targets were augmented to display 3D door/ window objects, as well as a series of buttons to provide access to object

Fig. 4. Wikitude door augmentation.

308

Automation in Construction 85 (2018) 305–316

M. Chu et al.

Fig. 5. Object test augmentations.

identical document set was provided to both groups. However, for the non-control group (BIM AR), the document set also included Quick Response (QR) markers on the door/window schedule page. In addition to the documents, participants were also provided with the mobile BIM AR system with access to the Wikitude application. Using the information contained within their document set, four tasks were evaluated:

via a mobile smartphone. The user logs in to the registered user account for the application and synchronizes data to the mobile version of the platform to auto detect 2D markers. The user can augment the 2D marker by holding the device with the marker in view, as illustrated in Fig. 5. No local information is required to be stored on the physical device itself, as an interface layer, containing buttons or text based links, can be embedded in the augmentation, so as to link the user to object-oriented, cloud-based building information. This, centralization of information in a single accessible location means the system is capable of being updated and adapted to changes in usage requirements with minimal effort.

1. For the doors listed below please state the number of hinges and the description as specified in the documents provided. 2. What is the fire rating and material finish specified for door D4? 3. Who are the acceptable manufacturers specified for W1 in Library A? 4. According to the specification what Australian Standard applies to the frame installation for W2?

3.2. Simulation Kerosuo et al. [11] found that models were often underutilized by construction teams and that the workers tended not to find them not particularly useful. This was due to BIM representations requiring mental effort to extract information effectively, and therefore workers preferred to use 2D paper plans to obtain information, as they required less effort. However, the process of retrieving information from 2D documentation can be a time-consuming as workers are dependent upon indexing to determine its location [29]. To evaluate how the mobile BIM AR system could effectively improve on this process, an experimental control group study was designed to simulate different scenarios for retrieving information. This research was developed using 2D paper based documentation, as paper-based systems are still considered by and large a universal format for information delivery within construction. It also enables a direct comparison of the working memory associated with the information retrieval process between traditional indexing and electronic search methods. A document set comprising 23 × A3 paper based plans and a binded 331 × A4 printed specification, which included a contents page, was used as the basis for evaluation. Experiment groups were established to provide comparison data between existing methods of information retrieval (control group, traditional 2D paper based documents) and the designed BIM AR system (non-control group). An

The information required to complete these tasks was located in separate sections of the specification document. Participants were required to familiarise themselves with the format and structure of the documents provided. The identical tasks were presented to both the control and non-control group to enable a comparative evaluation between existing and the mobile BIM AR methods for information retrieval to be undertaken. The integration of the markers into the drawings enables the evaluation of the quality of the interaction between the individual and the BIM AR system to be determined [29]. This does not, however, establish a basis to resolve issues pertaining to the positioning of markers in a construction environment [18,39,40].

3.3. Experimental control group studies To evaluate the performance of the information retrieval processes, it was necessary to select a population group that was able to understand typical building document sets. Thus, purposive sampling was used for the recruitment of participants for this research. The population group for this research comprised of 20 academic staff and final year undergraduate construction management students, all of whom had a minimum of one year experience working within the construction 309

Automation in Construction 85 (2018) 305–316

M. Chu et al.

Participant responses on the answer sheet were assessed. Each task answer was given a score of one, if the participant had retrieved the correct information, with a maximum score of four, if all tasks were correctly completed. This allowed the accuracy of the information attained through the two methods to be compared. The results were then analysed using descriptive statistics to provide an overview of performance for each of the groups. The ‘Post Experiment Questionnaire’ was adapted from Olsson [41], which contained subjective measures that focused on the quality of human interaction with technology (Table 1). The questionnaire was identical for both groups. The intent of the questionnaire was to identify the factors influencing the quality of interaction and the capabilities of the two methods in assisting in the completion of the experiment tasks. (Adapted from [41]). To beta-test the developed mobile BIM AR system and the experiment design, a pilot study was carried out using a volunteer to trial the camera functionalities, target detection and access to related

industry. A breakdown of the samples characteristics is presented in Table 2. Each participant was randomly allocated to undertake the experiment using either the 2D Paper based medium (control), or 2D Paper based medium supported by BIM AR (non-control) to perform the required tasks. This resulted in a sample size of 10 participants in each group. The sample size is similar to Wang and Dunston's [32] research who relied on 16 graduate engineering students. 3.3.1. Procedure An information and blank answer sheet was prepared for participants, which explained the experiment's requirements and detailed the four tasks to be undertaken. Participants completed the experiment individually. The researcher observed each individual's performance and recorded their completion times for each of the tasks. Directly after completing the four tasks each participant was asked to complete a post experiment questionnaire. Table 1 Questionnaire. Subjective measures

Questions

Characterization of User

Question 1: Did you use the Augmented Reality system or 2D paper drawings today (Circle one)? Augmented Reality/2D paper drawings Question 2: What is your gender? (Selection Response) Male/Female/Rather not identify Question 3: What is your age? (Statement Response) ___________/Rather not state Question 4: How many years' experience do you have within the construction industry? (Selection Response) □ 0–5 years □ 5–10 years □ 10–20 years □ Over 20 years Question 5: What level in a project team do you generally consider yourself? (Selection Response) □ □ □ □ □ □ □

Project Management Cost Management Contract administrator Architect/Designer Engineer/Technical Resource On-site construction personnel Other (please state) _______________

Question 6: Did you know what Building Information Modelling (BIM) was prior to today's experiment? (Selection Response) Yes/No Question 7: How confident are you generally with utilizing the information within Models? (Selection Response) □ Very confident □ Confident □ Somewhat confident □ Not very confident □ Not confident at all □ Unable to answer Question 8: In your day to day activities, how do you primarily access building information data (tick one) (Selection Response) □ Electronic 2D/3D CAD drawings (incl. PDF's of drawing sets) □ Native building models (.RVT, .DWG, etc.) □ Printed drawings and other documentation Question 9: What type of systems or technologies do you currently use to access building information? (tick all that apply) (Selection Response) □ □ □ □ □

Printed drawings Tablet Smartphone Laptop PC Desktop PC (continued on next page)

310

Automation in Construction 85 (2018) 305–316

M. Chu et al.

Table 1 (continued) Subjective measures

Questions

Extent of the experience felt

Question 10: Please rate the following in order of importance to you from 1 to 5 (1 = most important, 5 = least important). (Statement Response) ( ) Portability and ease of access to relevant project information on-site ( ) Timeliness of revisions during the construction period ( ) Clarity of building information between different stakeholders of a project ( ) Reducing errors and omissions of documentation ( ) Clear communication between project team members Question 11: How difficult was the information to find? (Selection response) Easy/Medium/Hard/Couldn't locate it Question 12: Did you feel that the tools given to you were sufficient to complete the required tasks? (Selection Response) Yes/No Question 13: If you used the AR system today, what processes do you think would benefit from future application of AR technology? (Selection Response)

User opinions of the quality of interaction

□ Design and layout planning □ As-planned vs. As-built determination □ Spatial collision detection □ Progress monitoring □ Safety Training □ Collaborative communication □ Construction education □ None at all □ Did not use the AR system Question 14: Please comment on aspects that made it difficult for you to complete the tasks? (Statement Response) Question 15: Please comment on how the documents or tools you used today helped you locate the information for completion of the tasks. (Statement Response)

construction industry in Western Australia was in a ‘boom’ period when the students commenced their studies in 2012, with all being offered part-time work by the end of their first year. By the time the students had entered their final year they had acquired considerable experience in a variety of roles (Table 2). Several large-scale infrastructure projects have mandated BIM to Level of Development 500 (i.e. The ‘Model Element’ is a field verified representation in terms of size, shape, location, quantity, and orientation), which a some of the student participants had been involved with delivering. The preferred medium for data retrieval varied between groups, but surprisingly printed drawings and documents were identified as a preferred method by 4 (20%) of participants. Mobile devices such as smart phones and tablets were identified as the least preferred options for retrieving information; an issue here was the difficultly associated with deciphering the text displayed on the screen. Table 3 presents a summary of the participants' experience of using the 2D paper based documents (control group) or the BIM AR system (non-control group). The information retrieval process using the BIM AR system was generally identified as being ‘easy’, while the control group found retrieving information a more difficult (‘medium’ difficulty) (Table 3). Those that had used the BIM AR system were asked what application AR might have in the future. The potential for collaborative communication as well as the ability to compare ‘as-planned’ with ‘as-built’ were recognized as being applications that could potentially improve the efficiency and effectiveness of tasks during construction. A summary of the quality of interactions between 2D paper based documents and BIM AR system are presented in Table 4.

documents located in the cloud. Multiple markers were scanned in succession, while making changes to the base information stored on the Wikitude database and the Microsoft OneDrive, in order to determine if the mobile application system could robustly deal with information updates. The procedure for carrying out the experiment was as follows:

• Introduction: The aim of the evaluation was explained to partici-

• • •

pants. For both experiment groups, the tools, documents and the BIM AR (where applicable) were described/demonstrated. An explanation of how the documents and systems linked with one another and the four tasks, as listed on the ‘Experiment Answer Sheet’, was provided to participants; Familiarisation of tools and documents: The participants were all given the opportunity to take time to familiarise themselves with the provided experiment medium tools and documents. Experiment process: Once familiarised with the tools and documents, the participants commenced the tasks. The participants were able to freely carry out their information searching document. The completion times for each task, as well as any notable observations were recorded for each individual. Post Experiment Questionnaire: Directly after finishing the tasks the participants were asked to complete the post experiment questionnaire in order to capture their experience and opinions relating to the use of the tools and documents provided in the experiment.

4. Results and discussion 4.1. Characteristics, experience and quality of interactions

4.2. Retrieval of information

Table 2 presents the characteristics of the participants. The sample comprised 18 males and 2 females. As the use of AR has been demonstrated as equally effective for working memory between genders [42], different genders were not analysed. The mean age was approximately 26 years for both groups. The age range and years of experience working in construction industry were also similar. Noteworthy, the

The completion times recorded for each task for both control groups were analysed to compare the performance of the two different information retrieval mediums used in the evaluation. Summary statistics are given for the four experiment tasks, and are presented in Tables 5 and 6 for the control and non-control groups, respectively. Overall task completion times are then summarized in Table 7. In addition, 311

Automation in Construction 85 (2018) 305–316

M. Chu et al.

Table 2 Summary of users characteristic.

Table 3 Summary of participants' experiences. Qualitative Responses (out of 10 per group) Control group

Qualitative Responses (out of 10 per group) Non-control group

Control group

Characterization questions

Participant experience

Age (years) Range: Mean:

21–38 26.3

21–36 26.8

Experience (years) a) 0–5: b) 6–10: c) 11–20: d) > 21:

8 N/A 2 N/A

7 2 1 N/A

Discipline a) Project management: b) Cost management: c) Contract administrator: d) Architect/designer: e) Engineer/technical: resource: f) On-site construction personnel: g) Other – data management: h) Other – student:

4 2 N/A N/A 2 1 N/A 1

1 3 N/A 1 1 N/A 1 1

Inconclusive

Inconclusive

Experience with BIM a) Yes: b) No:

10 N/A

10 N/A

BIM utilization level a) Very confident: b) Confident: c) Somewhat confident: d) Not very confident: e) Not confident at all: f) Unable to answer:

1 3 3 2 N/A 1

3 2 5 N/A N/A N/A

Preferred data retrieval method a) Electronic 2D/3D CAD & PDF'S b) Native models.(rvt, .dwg, etc.): c) Printed drawings/documents

8 1 1

4 3 3

Please rate the following in order of importance to you from 1 to 5

Non-control group

Technologies for data retrieval (multiple responses allowed) a) Printed drawings: 8/10 b) Tablet: 2/10 c) Smartphone: 2/10 d) Laptop PC: 9/10 e) Desktop PC: 8/10

How difficult was the information to find? Easy: N/A Medium: 6 Hard: 4 Could not locate it: N/A

9 1 N/A N/A

Did you feel that the tools given to you were sufficient to complete the required tasks? Yes: 10 10 No: N/A N/A If you used the AR system today, what processes do you think would benefit from future application of AR technology? (multiple responses allowed) a) Design and layout planning N/A 6/10 b) As-planned vs. As-built N/A 8/10 determination c) Spatial collision detection N/A 5/10 d) Progress monitoring N/A 7/10 e) Safety Training N/A 5/10 f) Collaborative communication N/A 9/10 g) Construction education N/A 7/10 h) None at all N/A N/A i) Did not use the AR system 10/10 N/A j) Other – maintenance N/A 1/10

which were proportioned as 1 mark per individual task, with partial marks also being awarded. The time to retrieve information using the mobile BIM AR system were reflected across all task completion periods, as denoted in Fig. 6, with the exception of Task 4 where the control group recorded faster times on average. However, their faster recorded times can be attributed to the location of the information relating to Task 4 in the specification book, which was located close the documentation required for Task 3. The slower completion times for the control group, indicates that information retrieval performance is influenced by the person's ability to familiarise themselves with the medium itself, during the time allocated for interpreting the information. It was observed that the control group participants at the commencement of the experiment appeared to be uncertain of where to locate the information in the 2D documentation due to its layout and structure. The experiment demonstrated the capability of the BIM AR system to reduce learning curve times for new processes in comparison to paper manuals. In this instance, participants' memory capacity and cognitive load were lowered, which aligns with the research reported in Hou et al. [12]. Slower times for the control group were attributable to nature and volume of information contained in 331-page specification document. Participants indicated that they found it difficult to visually identify the information required to complete the task. This was despite the document containing a contents page that was linked to all the sections referenced in the experiment task sheet. It was however, often underutilized, as participants appeared to skim through the documents pages and attempted to identify relevant sections using keyword identifiers. Despite many of the participants being able to navigate to the correct sections, they often became lost when attempting to verify and confirm the information that was sought; the documentation simply contained too much information. Essentially, the control group engaged in a cyclical process of familiarising themselves with the document, cross-referencing to determine keyword matches, and attempted to verify the source data. If unsuccessful, they repeated this sequence until information was found. Naturally, this increased the time to conduct a task and the propensity for errors to be made (Fig. 7).

6/10 3/10 3/10 10/10 7/10

variances between the two mediums used in the evaluation are calculated and provided (Tables 5 to 7). These variances are displayed as faster times represented by (−) and slower times represented by (+), from the perspective of 2D compared to the mobile BIM AR system that was developed. The results of the experiments demonstrated that the group of participants using the mobile BIM AR system were able to complete their respective four tasks in approximately 1 h 42 min compared to their paper based documents only counterparts, who took a total time of 3 h 08. The faster results by the non-control group using the mobile BIM AR system, appears to be due to the systems overall ability to support the retrieval of information (Fig. 6). Hence, they were able to complete their tasks generally more efficiently and effectively, when compared to the control group who relied on existing manual processes, which involved searching, indexing and retrieving [29]. The experiment answer sheets completed during the evaluation were assessed to determine the accuracy of the respective information retrieval tasks (Table 8). Available marks were allocated for each group, 312

Automation in Construction 85 (2018) 305–316

M. Chu et al.

Table 4 User opinions of the quality of interaction. Thematic Responses Control group

Non-control group

User opinions of the quality of interaction Difficulties encountered that made it difficult for you to complete the tasks?

a) Information Volume - (4/10) b) Issues with the medium's structure and layout - (6/10)

Please comment on how the documents or tools you used today helped you locate the information for completion of the tasks.

a) Use of Keywords and searching using indexing - (7/10) b) Satisfied/could be improved - (3/10)

a) Marker registration and speed of database (3/10) b) Non-clarity of information in Database - (2/10) c) Functionality and usability of the mobile application a) Fast access to information – (4/10) b) Reduced mental effort/able to find the right information - (6/10)

Table 5 Descriptive statistics for task completion times Task 1 and 2. Task 1

Mean Median Mode Range Min. Max. Total Count

Task 2

2D Paper

BIM AR

2D vs. BIM AR

2D Paper

BIM AR

2D vs. BIM AR

0:10:25.09 0:10:02.84 N/A 0:06:10.36 0:07:54.02 0:14:04.38 1:44:10.86 10

0:05:15.23 0:05:05.66 N/A 0:05:59.21 0:03:01.66 0:09:00.87 0:52:32.28 10

+ 0:05:09.86 + 0:04:57.18

0:04:44.63 0:04:50.41 N/A 0:04:28.28 0:02:40.20 0:07:08.48 0:47:26.31 10

0:02:35.80 0:02:12.15 N/A 0:03:29.68 0:01:21.66 0:04:51.34 0:25:58.03 10

+ 0:02:08.83 + 0:02:38.25

+ 0:00:11.15 + 0:04:52.36 + 0:05:03.51 + 0:51:38.58

+ 0:00:58.60 + 0:01:18.54 + 0:02:17.14 + 0:21:28.28

Table 6 Descriptive statistics for task completion times Tasks 3 and 4. Task 3

Mean Median Mode Range Min. Max. Total Count

Task 4

2D Paper

BIM AR

2D vs. BIM AR

2D Paper

BIM AR

2D vs. BIM AR

0:02:47.15 0:01:51.88 N/A 0:07:16.25 0:00:31.61 0:07:47.86 0:27:51.48 10

0:01:21.55 0:01:11.08 N/A 0:01:27.79 0:00:49.35 0:02:17.14 0:13:35.52 10

+ 0:01:25.60 + 0:00:40.80

0:00:55.51 0:00:56.83 N/A 0:01:13.34 0:00:12.14 0:01:25.48 0:09:15.08 10

0:01:01.61 0:00:53.04 N/A 0:01:36.06 0:00:45.31 0:02:21.37 0:10:16.14 10

− 0:00:06.11 + 0:00:03.79

+ 0:05:48.46 + 0:00:17.74 + 0:05:30.72 + 0:14:15.96

− 0:00:22.72 − 0:00:33.17 − 0:00:55.89 − 0:01:01.06

addition, the ease of access resulted in a reduction in errors committed (2.3 points or a 6% variance) when compared to the control group. The result could be significant when time is considered, as those attained were not only more accurate than the control group, but were achieved in under half the time required for the manual methods. The findings presented validate the research reported in Hou et al. [12], with respect to improved mental workloads. However, this research has revealed that lower mental workload was associated with the QR markers ability to augment and link the participants to sources of data; this reduced mental workload requirements to process the

As participants became more familiar with the document, the time taken to complete subsequent tasks reduced rapidly, as reflected as they became dependent on their working memory Hou et al. [12]. Given the age range (21 to 38 years) of participants for the control group, this finding may be associated with the characteristics of sample, as 8/10 of the control group's participants indicated the they preferred accessing information using an electronic medium (Table 2). In comparison, the non-control group using the mobile BIM AR system were more effective and efficient due to the systems capabilities to provide a direct path for information retrieval, which required reduced mental effort. In

Table 7 Overall task times for control groups.

Control group Non-control group 2D vs. BIM AR Variance (%)

Task 1 Total

Task 2 Total

Task 3 Total

Task 4 Total

Group total Time

1:44:10.86 0:52:32.28 +0:51:38.58 +49.57%

0:47:26.31 0:25:58.03 + 0:21:28.28 + 45.26%

0:27:51.48 0:13:35.52 +0:14:15.96 +51.21%

0:09:15.08 0:10:16.14 − 0:01:01.06 − 11.00%

3:08:43.73 1:42:21.97 +1:26:21.76 +45.76%

313

Automation in Construction 85 (2018) 305–316

M. Chu et al.

Fig. 6. Overall task times for control and non-control groups.

1:55:12.00 1:40:48.00 1:26:24.00

Time

1:12:00.00 0:57:36.00 0:43:12.00 0:28:48.00 0:14:24.00 0:00:00.00 Control Group Non-Control Group

Task 1 1:44:10.86 0:52:32.28

Task 2 0:47:26.31 0:25:58.03

Task 3 0:27:51.48 0:13:35.52

Task 4 0:09:15.08 0:10:16.14

Task ID

Table 8 Information retrieval medium accuracy.

Control group Non-control group

Task 1

Task 2

Task 3

Task 4

Total

Accuracy

7.7 10

9 9.5

9 10

10 8.5

35.7 38

89% 95%

information. 5. Research limitations The strength of an experiment lays in its capacity to demonstrate cause-and-effect relationships. In this research, this relationship focused on the efficacy to retrieve of information and the use of AR-BIM tool. To establish a cause-and-effect, that is, task efficiency improves when using an AR-BIM tool to retrieve information, the researchers constructed a scenario to determine this relationship. While such an approach can provide internal validity (i.e. approximate truth about inferences regarding cause-and-effect), generalizations (external validity) may not be concluded as the results are derived from an artificial situation that removed from the real-world. Previous AR research undertaken in construction has enabled the process of replication to be undertaken to examine task efficiency (e.g., [31,42]). It should be acknowledged that the research has limitations; the sample is small, but comparable to other studies of this nature (e.g., [12,31,32,42]). In addition, the evaluation of the BIM AR system was limited to four tasks and for individual participants, the overall experiment times rarely exceeded 20 min. Other limitations of this research that need to be identified pertain to the demographics of the sample, namely age and gender. However, most research studies examining the usability of AR have been undertaken with young adults and the ratio of male to female being low [43].

Fig. 7. Mental data verification flow.

This was found to be due to the systems capabilities to provide the participants with direct paths to information sources. As a result, this lowered the non-control groups mental workloads and enabled tasks to be completed productively and with minimal error. The research also revealed that information volume, as well as the structure of 2D paper documentation influenced a person's ability to familiarise themselves with the medium and effectively extract information. Significantly slower times and higher rates of error occurred with the control group as they were unable familiarise themselves with the structure of the documentation that was provided. This research presented in this paper has demonstrated that a minor modification to existing information formats (2D plans) with the inclusion of Quick

6. Conclusion Research was undertaken to evaluate the effectiveness of BIM and AR system integration to enhance task productivity through improved information handling. The findings of the evaluation undertaken for the designed and developed cloud-based mobile BIM AR system improved the users' capabilities in undertaking information retrieval processes. 314

Automation in Construction 85 (2018) 305–316

M. Chu et al.

[12] L. Hou, X. Wang, L. Bernold, P.E.D. Love, Using animated augmented reality to cognitively guide assembly, ASCE J. Comput. Civ. Eng. 27 (5) (2013) 439–451, http://dx.doi.org/10.1061/(asce)cp.1943-5487.0000184. [13] S. Meža, Z. Turk, M. Dolenc, Component based engineering of a mobile BIM-based Augmented Reality system, Autom. Constr. 42 (2014) 1–12, http://dx.doi.org/10. 1016/j.autcon.2014.02.011. [14] X. Wang, S.K. Ong, A.Y.C. Nee, A comprehensive survey of augmented reality assemnbly research, Adv. Manuf. 4 (1) (2016) 1–22, http://dx.doi.org/10.1007/ s4043. [15] X. Wang, M.J. Kim, P.E.D. Love, S.-C. Kang, Augmented reality in built environment: classification and implications for future research, Autom. Constr. 32 (2013) 1–13, http://dx.doi.org/10.1016/j.autcon.2012.11.021. [16] A.H. Behzadan, Z. Aziz, C.J. Anumba, V.R. Kamat, Ubiquitous location tracking for context-specific information delivery on construction sites, Autom. Constr. 17 (6) (2008) 737–748, http://dx.doi.org/10.1016/j.autcon.2008.02.002. [17] Construction Industry Council, Built Environment 2050: A Report on our Digital Future. BIM 2050, (2014) (Available at: be2050-cic-bim2050-2014-1.pdf, Accessed 21st October 2016). [18] J. Irizarry, M. Gheisari, G. Williams, B.N. Walker, Infospot: a mobile augmented reality method for accessing building information through a situation awareness approach, Autom. Constr. 33 (2013) 11–23, http://dx.doi.org/10.1016/j.autcon. 2012.09.002. [19] Y. Arayici, T. Onyenobi, C. Egbu, Building information modelling for facilities management: the mediacity case study approach, Int. J. 3-D Inf. Model. (IJ3DIM) 1 (1) (2012) 55–73, http://dx.doi.org/10.4018/ij3dim.2012010104. [20] L.J. Chen, H. Luo, A BIM-based construction quality management model and its applications, Autom. Constr. 46 (2014) 64–73, http://dx.doi.org/10.1016/j.autcon. 2014.05.009. [21] R. Volk, J. Stengel, F. Schultmann, Building information modeling for existing buildings - literature review and future needs, Autom. Constr. 38 (2014) 109–127, http://dx.doi.org/10.1016/j.autcon.2013.10.023. [22] M. Golparvar-Fard, F. Pena-Mora, S. Savarese, Integrated sequential ‘as-built’ and ‘as-planned’ representation with D(4)AR tools in support of decision-making tasks in the AEC/FM industry, ASCE J. Constr. Eng. Manag. 137 (12) (2011) 1099–1116, http://dx.doi.org/10.1061/(asce)co.1943-7862.0000371. [23] S.G. Naoum, Factors influencing labor productivity on construction sites: a state-ofthe-art literature review and a survey, Int. J. Product. Perform. Manag. 65 (3) (2016) 401–421, http://dx.doi.org/10.1108/IJPPM-03-2015-0045. [24] E. Brynjolfsson, The productivity paradox of information technology, Communun. ACM 36 (12) (1993) 66–77, http://dx.doi.org/10.1145/163298.163309. [25] P. Karr-Wisniewski, Y. Lu, When more is too much: operationalizing technology overload and exploring its impact on knowledge worker productivity, Comput. Hum. Behav. 26 (5) (2010) 1061–1072, http://dx.doi.org/10.1016/j.chb.2010.03. 008. [26] A. Fazli, S. Fathi, M.H. Enferadi, M. Fazli, B. Fathi, Appraising effectiveness of building information management in project management, Procedia Technol. 16 (2014) 1116–1125, http://dx.doi.org/10.1016/j.protcy.2014.10.126. [27] M. Loosemore, Impediments to reform in the Australian building and construction industry, Aust. J. Constr. Econ. Build. 3 (2) (2012), http://dx.doi.org/10.5130/ AJCEB.v3i2.2913. [28] P.E.D. Love, D.J. Edwards, S. Han, Y.M. Goh, Design error reduction: toward the effective utilization of building information modelling, Res. Eng. Des. 22 (3) (2011) 173–187, http://dx.doi.org/10.1007/s00163-011-0105-x. [29] X. Wang, M. Truijens, L. Hou, Y. Wang, Y. Zhou, Integrating augmented reality with building information modeling: onsite construction process controlling for the liquefied natural gas industry, Autom. Constr. 40 (2014) 96–105, http://dx.doi.org/ 10.1016/j.autcon.2013.12.003. [30] P.S. Dunston, X. Wang, Mixed reality-based visualization interfaces for architecture, engineering, and construction industry, ASCE J. Constr. Eng. Manag. 131 (12) (2005) 1301–1309, http://dx.doi.org/10.1061/(asce)0733-9346(2005) 131:12(1301). [31] L. Hou, X. Wang, M. Truijens, Using augmented reality to facilitate piping assembly: an experiment-based evaluation, ASCE J. Comput. Civ. Eng. 29 (1) (2015), http:// dx.doi.org/10.1061/(ASCE)CP.1943-5487.0000344, 05014007. [32] X. Wang, P.S. Dunston, Compatibility issues in augmented reality systems for AEC: an experimental prototype study, Autom. Constr. 15 (3) (2006) 314–326, http://dx. doi.org/10.1016/j.autcon.2005.06.002. [33] M. Nakanishi, T. Sato, Application of digital manuals with a retinal imaging display in manufacturing: behavioral, physiological, and psychological effects on workers, Hum. Factors and Ergon. Manuf Serv. Ind. 25 (2) (2015) 228–238, http://dx.doi. org/10.1002/hfm.20535. [34] H.-L. Chi, S.C. Kang, X. Wang, Research trends and opportunities of augmented reality applications in architecture, engineering, and construction, Autom. Constr. 33 (2013) 116–122, http://dx.doi.org/10.1016/j.autcon.2012.12.017. [35] K. Peffers, T. Tuunanen, M.A. Rothenberger, S. Chatterjee, A design science research methodology for information systems research, J. Manag. Inf. Syst. 24 (3) (2007) 45–77, http://dx.doi.org/10.2753/MIS0742-1222240302. [36] G.L. Geerts, A design science research methodology and its application to accounting information systems research, Int. J. Account. Inf. Syst. 12 (2) (2011) 142–151, http://dx.doi.org/10.1016/j.accinf.2011.02.004. [37] J.E. Van Aken, Management research as a design science: articulating the research products of mode 2 knowledge production in management, Br. J. Manag. 16 (1) (2005) 19–36, http://dx.doi.org/10.1111/j.1467-8551.2005.00437.x. [38] goQR, Qr Code Generator, Available at: http://goqr.me/, (2016) , Accessed date: 11 February 2016. [39] S. Lee, O. Akin, Augmented reality-based computational fieldwork support for

Response marker can significantly improve the information retrieval process as well as potential for BIM and AR integration to enhance productivity. In consideration of the limited number of tasks undertaken by participants and the overall experimental period, future research should focus on whether the BIM AR are system is useful for an extended duration after familiarisation with the medium is achieved, or the attainment of a learning curve for a new process. In addition, task design during construction should also be integrated with the building information model to evaluate their effectiveness, as the data extracts used in this research for the mobile BIM AR system were non-complex and text based in nature. As the construction industry embraces BIM there is a perception that task productivity will improve in construction. However, no empirical evidence exists to support this view. The implementation of BIM exposes workers on-site with increasing amounts of information, which can hinder rather than improve decision-making and productivity. Existing AR software can be readily customized, at a low cost, and adapted to accommodate a multitude of tasks in construction. While the integration of BM and AR offers the potential to improve information extraction on-site, it can also reduce the propensity of workers to commit cognitive failures. Consequently, errors that arise due to lapses, slips and mistakes, which can lead to rework being undertaken can be mitigated as workers' mental work load is reduced. The integration of BIM and AR in construction is an area of research that has received limited attention. As AR technology matures, new opportunities for its application to construction emerge, with the benefits to be gained from its implementation outweighing the cost. Acknowledgments The authors would like to thank the anonymous reviewers for their constructive comments, which enabled the quality of this manuscript to be improved. The authors would like to thank the staff and students from the Department of Construction Management at Curtin University in Australia who participated in this research. References [1] S. Changali, A. Mohammad, M. van Nieuwland, The construction productivity imperative. McKinsey & Company, Available at http://www.mckinsey.com/ industries/capital-projects-and-infrastructure/our-insights/the-constructionproductivity-imperative:, (2015) , Accessed date: 17 November 2016. [2] A. Sanchez, K. Hampson, S. Vaux, Delivering Value with BIM: A Whole-of-Life Approach, Routledge, London, 2016 (ISBN-13: 978-1138118997). [3] K. Kim, Y. Cho, S. Zhang, Integrating work sequences and temporary structures into safety planning; automated scaffolding-related safety hazard and prevention in BIM, Autom. Constr. 70 (2016) 128–142, http://dx.doi.org/10.1016/j.autcon.2016. 06.012. [4] T. Nath, M. Attarzadeh, R.L.K. Tiong, Precast workflow productivity measurement through BIM, Proceedings of the Institution of Civil Engineers – Management, Procurement and Law, 169(5) 2016, pp. 208–216, , http://dx.doi.org/10.1680/ jmapl.15.00045. [5] P.E.D. Love, I. Simpson, A. Hill, C. Standing, From justification to evaluation: building information Modelling for asset owners, Autom. Constr. 35 (2013) 208–216, http://dx.doi.org/10.1016/j.autcon.2013.05.008. [6] P.E.D. Love, I. Simpson, A. Hill, J. Matthews, O. Olatunji, Benefits realization management of building information Modelling: obtaining value for asset owners, Autom. Constr. 37 (2014) 1–10, http://dx.doi.org/10.1016/j.autcon.2013.09.007. [7] J. Matthews, P.E.D. Love, S. Heinemann, C. Rumsey, R. Chandler, O. Olatunji, Realtime progress monitoring: Process re-engineering with cloud-based BIM in construction, Autom. Constr. 58 (2015) 38–47, http://dx.doi.org/10.1016/j.autcon. 2015.07.004. [8] C. Agarwala, Technology and knowledge worker productivity, Int. J. Comput. Appl. 102 (1) (2014), http://dx.doi.org/10.5120/17783-8564. [9] C.J. Anumba, J. Pan, R.R.A. Issa, I. Mutis, Collaborative project information management in a semantic web environment, Eng. Constr. Archit. Manag. 15 (1) (2008) 78–94, http://dx.doi.org/10.1108/09699980810842089. [10] S. Aral, E. Brynjolfsson, M. Van Alstyne, Information, technology, and information worker productivity, Inf. Syst. Res. 23 (3) (2012) 849–867, http://dx.doi.org/10. 1287/isre.1110.0408. [11] H. Kerosuo, R. Miettinen, S. Paavola, T. Mäki, J. Korpela, Challenges of the expansive use of Building Information Modeling in construction projects, Production 25 (2015) 289–297, http://dx.doi.org/10.1590/0103-6513.106512.

315

Automation in Construction 85 (2018) 305–316

M. Chu et al.

SpringerLink, New York, 2013(978-1-4614-4204-2). [42] L. Hou, X. Wang, A study on the benefits of augmented reality in retaining working memory in assembly tasks: a focus on differences in gender, Autom. Constr. 32 (2013) 38–45, http://dx.doi.org/10.1016/j.autcon.2012.12.007. [43] A. Dey, M. Billinghurst, R.W. Lindeman, J.E. Swan II, Systematic review of usability studies in Augmented Reality between 2005 and 2015, Mixed and Augmented Reality (ISMAR-Adjunct), IEEE International Symposium, 2016, pp. 49–50, , http:// dx.doi.org/10.1109/ISMAR-Adjunct.2016.0036.

equipment operations and maintenance, Autom. Constr. 20 (4) (2011) 338–352, http://dx.doi.org/10.1016/j.autcon.2010.11.004. [40] S.K. Ong, M.L. Yuan, A.Y.C. Nee, Augmented reality applications in manufacturing: a survey, Int. J. Prod. Res. 46 (10) (2008) 2707–2742, http://dx.doi.org/10.1080/ 00207540601064773. [41] T. Olsson, Concepts and subjective measures for evaluating user experience of mobile augmented reality services, in: Weidong Huang, Leila Alem, Mark A. Livingston (Eds.), Human Factors in Augmented Reality Environments,

316